CentOS 6.4 died on me again. Didn't leave any traces that I could find. The screen just suddenly went black. Couldn't switch to another virtual terminal. Pushing the reset button worked. Didn't have to power off this time.
On Sun, 24 Nov 2013, Michael Hennebry wrote:
CentOS 6.4 died on me again. Didn't leave any traces that I could find. The screen just suddenly went black. Couldn't switch to another virtual terminal. Pushing the reset button worked. Didn't have to power off this time.
For some reason I got back an old set of Firefox tabs. The second to last time I started Firefox, it didn't give me my old tabs. It just welcomed me to CentOS.
On Sun, 24 Nov 2013, Michael Hennebry wrote:
On Sun, 24 Nov 2013, Michael Hennebry wrote:
CentOS 6.4 died on me again. Didn't leave any traces that I could find. The screen just suddenly went black. Couldn't switch to another virtual terminal. Pushing the reset button worked. Didn't have to power off this time.
For some reason I got back an old set of Firefox tabs. The second to last time I started Firefox, it didn't give me my old tabs. It just welcomed me to CentOS.
Oops. Never mind: I'd booted into Fedora 14 by mistake.
On Sun, 24 Nov 2013, John R Pierce wrote:
On 11/24/2013 9:45 PM, Michael Hennebry wrote:
CentOS 6.4 died on me again.
only time that has EVER happened to me, on dozens and dozens of systems, has been when there's been a serious hardware problem.
I really do not know whether to hope you are correct. On one hand a new computer would be expensive. On the other, if it's something else, my diagnostic skills are clearly not up to the task.
On Mon, Nov 25, 2013 at 10:25 AM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
On Sun, 24 Nov 2013, John R Pierce wrote:
On 11/24/2013 9:45 PM, Michael Hennebry wrote:
CentOS 6.4 died on me again.
only time that has EVER happened to me, on dozens and dozens of systems, has been when there's been a serious hardware problem.
I really do not know whether to hope you are correct. On one hand a new computer would be expensive. On the other, if it's something else, my diagnostic skills are clearly not up to the task.
Keep an eagle eye on dmesg and the logs. If you can, bring machine down and run memtest86 for a few hours (say, when you go to bed or is out partying). Also, *sometimes* the messages log might say something interesting. But I would start with dmesg.
There are some HD tests you can make but honestly I can't pull them off the fuzzy mist that is my head. Hardware or software raid?
-- Michael hennebry@web.cs.ndsu.NoDak.edu "On Monday, I'm gonna have to tell my kindergarten class, whom I teach not to run with scissors, that my fiance ran me through with a broadsword." -- Lily _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
On Mon, 25 Nov 2013, Mauricio Tavares wrote:
On Mon, Nov 25, 2013 at 10:25 AM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
On Sun, 24 Nov 2013, John R Pierce wrote:
On 11/24/2013 9:45 PM, Michael Hennebry wrote:
CentOS 6.4 died on me again.
only time that has EVER happened to me, on dozens and dozens of systems, has been when there's been a serious hardware problem.
I really do not know whether to hope you are correct. On one hand a new computer would be expensive. On the other, if it's something else, my diagnostic skills are clearly not up to the task.
Keep an eagle eye on dmesg and the logs. If you can, bringmachine down and run memtest86 for a few hours (say, when you go to
I've run the memory test that comes with the Fedora 13 install disk. My computer's memory got a clean bill of health. To me, neither dmesg nor Xorg.0.log says anything interesting.
bed or is out partying). Also, *sometimes* the messages log might say something interesting. But I would start with dmesg.
Thank you for the reminder. It does.
Nov 25 09:47:22 localhost abrtd: Sending an email... Nov 25 09:47:22 localhost abrtd: Email was sent to: root@localhost Nov 25 09:47:24 localhost abrtd: Duplicate: UUID Nov 25 09:47:24 localhost abrtd: DUP_OF_DIR: /var/spool/abrt/ccpp-2013-11-25-09:46:10-7871 Nov 25 09:47:24 localhost abrtd: Corrupted or bad directory '/var/spool/abrt/ccpp-2013-11-25-09:46:55-8008', deleting Nov 25 09:47:26 localhost abrtd: Directory 'ccpp-2013-11-25-09:47:25-8243' creation detected Nov 25 09:47:26 localhost abrt[8445]: Saved core dump of pid 8243 (/usr/bin/kdeinit4) to /var/spool/abrt/ccpp-2013-11-25-09:47:25-8243 (78938112 bytes) Nov 25 09:47:52 localhost abrtd: Sending an email... Nov 25 09:47:52 localhost abrtd: Email was sent to: root@localhost Nov 25 09:47:53 localhost abrtd: Duplicate: UUID Nov 25 09:47:53 localhost abrtd: DUP_OF_DIR: /var/spool/abrt/ccpp-2013-11-25-09:46:10-7871 Nov 25 09:47:53 localhost abrtd: Corrupted or bad directory '/var/spool/abrt/ccpp-2013-11-25-09:47:25-8243', deleting Nov 25 10:04:58 localhost ntpd[2077]: time reset +0.288044 s
I ran this for F in /dev/sd??* ; do ( tune2fs -l $F ; echo $F ) | grep -e dev -e UUID ; done | tee /tmp/tune2fs.txt to check for duplicate UUIDs. I used sort and my eyeballs to check. There weren't any. The hard drive in use is newer than the motherboard, but older than the video card. I zapped the first video card installing the new hard drive. The second one seemed to die on its own.
There are some HD tests you can make but honestly I can't pull them off the fuzzy mist that is my head. Hardware or software raid?
No raid.
On Mon, Nov 25, 2013 at 10:45 AM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
Keep an eagle eye on dmesg and the logs. If you can, bringmachine down and run memtest86 for a few hours (say, when you go to
I've run the memory test that comes with the Fedora 13 install disk. My computer's memory got a clean bill of health.
I've seen a machine where it took 3+ days of running memtest86 to catch the error. And then after replacing the RAM, the machine still crashed occasionally. Turned out the software RAID1 mirrors had mismatching contents caused by the bad RAM and even though it would check clean, sometimes the read would come from the other mirror. After fixing that, the server has run for years.
But in general, I always suspect power supplies first for mysterious crashes.
On Mon, Nov 25, 2013 at 1:54 PM, Les Mikesell lesmikesell@gmail.com wrote:
But in general, I always suspect power supplies first for mysterious crashes.
+1 power supplies with bad caps.
Two weeks ago I had one 2007 Antec EPS12V PS fail on me. Upon inspection, bad caps.. (2 of them) http://en.wikipedia.org/wiki/Capacitor_plague
YMMV
FC
On Mon, Nov 25, 2013 at 11:45 AM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
On Mon, 25 Nov 2013, Mauricio Tavares wrote:
On Mon, Nov 25, 2013 at 10:25 AM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
On Sun, 24 Nov 2013, John R Pierce wrote:
On 11/24/2013 9:45 PM, Michael Hennebry wrote:
CentOS 6.4 died on me again.
only time that has EVER happened to me, on dozens and dozens of systems, has been when there's been a serious hardware problem.
I really do not know whether to hope you are correct. On one hand a new computer would be expensive. On the other, if it's something else, my diagnostic skills are clearly not up to the task.
Keep an eagle eye on dmesg and the logs. If you can, bringmachine down and run memtest86 for a few hours (say, when you go to
I've run the memory test that comes with the Fedora 13 install disk. My computer's memory got a clean bill of health. To me, neither dmesg nor Xorg.0.log says anything interesting.
bed or is out partying). Also, *sometimes* the messages log might say something interesting. But I would start with dmesg.
Thank you for the reminder. It does.
Nov 25 09:47:22 localhost abrtd: Sending an email... Nov 25 09:47:22 localhost abrtd: Email was sent to: root@localhost Nov 25 09:47:24 localhost abrtd: Duplicate: UUID Nov 25 09:47:24 localhost abrtd: DUP_OF_DIR: /var/spool/abrt/ccpp-2013-11-25-09:46:10-7871 Nov 25 09:47:24 localhost abrtd: Corrupted or bad directory '/var/spool/abrt/ccpp-2013-11-25-09:46:55-8008', deleting Nov 25 09:47:26 localhost abrtd: Directory 'ccpp-2013-11-25-09:47:25-8243' creation detected Nov 25 09:47:26 localhost abrt[8445]: Saved core dump of pid 8243 (/usr/bin/kdeinit4) to /var/spool/abrt/ccpp-2013-11-25-09:47:25-8243 (78938112 bytes)
So abrt is having enough issues to spit out a core dump. Since it watches when other applications crash, it might be worth investigating that.
Nov 25 09:47:52 localhost abrtd: Sending an email... Nov 25 09:47:52 localhost abrtd: Email was sent to: root@localhost Nov 25 09:47:53 localhost abrtd: Duplicate: UUID Nov 25 09:47:53 localhost abrtd: DUP_OF_DIR: /var/spool/abrt/ccpp-2013-11-25-09:46:10-7871 Nov 25 09:47:53 localhost abrtd: Corrupted or bad directory '/var/spool/abrt/ccpp-2013-11-25-09:47:25-8243', deleting Nov 25 10:04:58 localhost ntpd[2077]: time reset +0.288044 s
I ran this for F in /dev/sd??* ; do ( tune2fs -l $F ; echo $F ) | grep -e dev -e UUID ; done | tee /tmp/tune2fs.txt to check for duplicate UUIDs. I used sort and my eyeballs to check. There weren't any. The hard drive in use is newer than the motherboard, but older than the video card. I zapped the first video card installing the new hard drive. The second one seemed to die on its own.
There are some HD tests you can make but honestly I can't pull them off the fuzzy mist that is my head. Hardware or software raid?
No raid.
K. Anything interesting from smartctl? Have you used bonnie++ before? I think if you run it in a window/screen and then keep an eye on dmesg you might find issues on the HD.
-- Michael hennebry@web.cs.ndsu.NoDak.edu "On Monday, I'm gonna have to tell my kindergarten class, whom I teach not to run with scissors, that my fiance ran me through with a broadsword." -- Lily _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
On Mon, 25 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
The computer is a DakTEch Freedom 4 P4 DDR System. The system board is a D865GBFL w/LAN,audio & video Processor Intel Pentium 4 3.2 Ghz 800FSB I got it in 2006.
<snip> It is getting long in the tooth. (I'm about to replace my m/b of the same vintage at home.)
Ouch. $200+ before installation, and I'd still have a computer some Linuxes don't like.
On Mon, 25 Nov 2013, Les Mikesell wrote:
On Mon, Nov 25, 2013 at 10:45 AM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
Keep an eagle eye on dmesg and the logs. If you can, bringmachine down and run memtest86 for a few hours (say, when you go to
I've run the memory test that comes with the Fedora 13 install disk. My computer's memory got a clean bill of health.
I've seen a machine where it took 3+ days of running memtest86 to catch the error. And then after replacing the RAM, the machine still
Clean bill from memtest86 running overnight. The memory is the newest part of the computer. After rebooting, it crapped out twice in half an hour. After that, I went back to F14.
On Mon, 25 Nov 2013, Mauricio Tavares wrote:
K. Anything interesting from smartctl? Have you used bonnie++before? I think if you run it in a window/screen and then keep an eye on dmesg you might find issues on the HD.
Hadn't heard of either until now. Looked up smartctl: [root@localhost ~]# smartctl -H /dev/sdb smartctl 5.40 2010-10-16 r3189 [i386-redhat-linux-gnu] (local build) Copyright (C) 2002-10 by Bruce Allen, http://smartmontools.sourceforge.net
=== START OF READ SMART DATA SECTION === SMART overall-health self-assessment test result: PASSED Please note the following marginal Attributes: ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE 190 Airflow_Temperature_Cel 0x0022 068 044 045 Old_age Always In_the_past 32 (0 34 33 25)
[root@localhost ~]# smartctl --test=short /dev/sdb smartctl 5.40 2010-10-16 r3189 [i386-redhat-linux-gnu] (local build) Copyright (C) 2002-10 by Bruce Allen, http://smartmontools.sourceforge.net
=== START OF OFFLINE IMMEDIATE AND SELF-TEST SECTION === Sending command: "Execute SMART Short self-test routine immediately in off-line mode". Drive command "Execute SMART Short self-test routine immediately in off-line mode" successful. Testing has begun. Please wait 1 minutes for test to complete. Test will complete after Tue Nov 26 10:02:10 2013
Use smartctl -X to abort test. [root@localhost ~]# smartctl -l selftest /dev/sdb smartctl 5.40 2010-10-16 r3189 [i386-redhat-linux-gnu] (local build) Copyright (C) 2002-10 by Bruce Allen, http://smartmontools.sourceforge.net
=== START OF READ SMART DATA SECTION === SMART Self-test log structure revision number 1 Num Test_Description Status Remaining LifeTime(hours) LBA_of_first_error # 1 Short offline Completed without error 00% 44279 - # 2 Short offline Completed without error 00% 44279 -
Michael Hennebry wrote:
On Mon, 25 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
The computer is a DakTEch Freedom 4 P4 DDR System. The system board is a D865GBFL w/LAN,audio & video Processor Intel Pentium 4 3.2 Ghz 800FSB I got it in 2006.
<snip> It is getting long in the tooth. (I'm about to replace my m/b of the same vintage at home.)
Ouch. $200+ before installation, and I'd still have a computer some Linuxes don't like.
Expensive board. Is this at home? I got a Gigabyte board and Core I-3 combo at MicroCenter for under $200, I think. "Installation"? Hey, we're sysadmins, we build our own systems! <g>
On Mon, 25 Nov 2013, Les Mikesell wrote:
On Mon, Nov 25, 2013 at 10:45 AM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
Keep an eagle eye on dmesg and the logs. If you can, bringmachine down and run memtest86 for a few hours (say, when you go to
I've run the memory test that comes with the Fedora 13 install disk. My computer's memory got a clean bill of health.
I've seen a machine where it took 3+ days of running memtest86 to catch the error. And then after replacing the RAM, the machine still
Clean bill from memtest86 running overnight. The memory is the newest part of the computer. After rebooting, it crapped out twice in half an hour. After that, I went back to F14.
<snip> If the memory looks good, and the disk looks good, the only two things left are either the CPU or the m/b. I don't remember from earlier in the thread - do you have a video card, or onboard video? If the former, that's the third second thing.... If that's not a hot card, you might consider replacing that first.
mark
On Tue, Nov 26, 2013 at 11:57:26AM -0500, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
On Mon, 25 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
The computer is a DakTEch Freedom 4 P4 DDR System. The system board is a D865GBFL w/LAN,audio & video Processor Intel Pentium 4 3.2 Ghz 800FSB I got it in 2006.
<snip> It is getting long in the tooth. (I'm about to replace my m/b of the same vintage at home.)
Ouch. $200+ before installation, and I'd still have a computer some Linuxes don't like.
Expensive board. Is this at home? I got a Gigabyte board and Core I-3 combo at MicroCenter for under $200, I think. "Installation"? Hey, we're sysadmins, we build our own systems! <g>
On Mon, 25 Nov 2013, Les Mikesell wrote:
On Mon, Nov 25, 2013 at 10:45 AM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
Keep an eagle eye on dmesg and the logs. If you can, bringmachine down and run memtest86 for a few hours (say, when you go to
I've run the memory test that comes with the Fedora 13 install disk. My computer's memory got a clean bill of health.
I've seen a machine where it took 3+ days of running memtest86 to catch the error. And then after replacing the RAM, the machine still
Clean bill from memtest86 running overnight. The memory is the newest part of the computer. After rebooting, it crapped out twice in half an hour. After that, I went back to F14.
<snip> If the memory looks good, and the disk looks good, the only two things left are either the CPU or the m/b. I don't remember from earlier in the
or the power supply...
thread - do you have a video card, or onboard video? If the former, that's the third second thing.... If that's not a hot card, you might consider replacing that first.
mark
On Tue, 26 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
Ouch. $200+ before installation, and I'd still have a computer some Linuxes don't like.
Expensive board. Is this at home? I got a Gigabyte board and Core I-3
Yes. It's the only computer I've owned with either a hard drive or more than 8K of memory.
I've used others.
combo at MicroCenter for under $200, I think. "Installation"? Hey, we're sysadmins, we build our own systems! <g>
What's this "we"? I've opened the case three times and done four things: I've installed a new hard drive. The old one is still working. I've zapped the original video card, AGP. I've blown dust. I've had someone else install new RAM.
When I get up the nerve to open it again, I'll blow dust and look for misplaced intestines.
In the mean time, I'll boot back to CentOS and see how long it lasts.
On Tue, 2013-11-26 at 11:34 -0600, Michael Hennebry wrote:
On Tue, 26 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
Ouch. $200+ before installation, and I'd still have a computer some Linuxes don't like.
Expensive board. Is this at home? I got a Gigabyte board and Core I-3
Yes. It's the only computer I've owned with either a hard drive or more than 8K of memory.
I've used others.
combo at MicroCenter for under $200, I think. "Installation"? Hey, we're sysadmins, we build our own systems! <g>
What's this "we"? I've opened the case three times and done four things: I've installed a new hard drive. The old one is still working. I've zapped the original video card, AGP. I've blown dust. I've had someone else install new RAM.
When I get up the nerve to open it again, I'll blow dust and look for misplaced intestines.
In the mean time, I'll boot back to CentOS and see how long it lasts.
You may want to check the bios for some settings. I recently fixed some issues with my server system here at home (gen 1 i7 920) by turning ACPI 2.0 off. I first replaced the power supply but that did improve the situation from disk errors once every day or 2 to once every 5 days. Make sure you have the latest bios as well.
Michael Hennebry wrote:
On Mon, 25 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
<snip>
On Mon, 25 Nov 2013, Les Mikesell wrote:
On Mon, Nov 25, 2013 at 10:45 AM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
Keep an eagle eye on dmesg and the logs. If you can, bringmachine down and run memtest86 for a few hours (say, when you go to
I've run the memory test that comes with the Fedora 13 install disk. My computer's memory got a clean bill of health.
I've seen a machine where it took 3+ days of running memtest86 to catch the error. And then after replacing the RAM, the machine still
Clean bill from memtest86 running overnight. The memory is the newest part of the computer. After rebooting, it crapped out twice in half an hour. After that, I went back to F14.
Oh, right, *just* as I hit <send>, I realized one more thing: you have checked both ~/.xsession-errors and /var/log/Xorg.0.log, right?
mark
On Tue, 26 Nov 2013, m.roth@5-cent.us wrote:
Oh, right, *just* as I hit <send>, I realized one more thing: you have checked both ~/.xsession-errors and /var/log/Xorg.0.log, right?
I'd looked at .../Xorg.0.log , but until now, not ~/.xsession-errors : GNOME_KEYRING_SOCKET=/tmp/keyring-BWwFtW/socket SSH_AUTH_SOCK=/tmp/keyring-BWwFtW/socket.ssh ** Message: NumLock remembering disabled because hostname is set to "localhost" Failed to play sound: File or data not found
(polkit-gnome-authentication-agent-1:2815): GLib-GObject-WARNING **: cannot register existing type `_PolkitError'
(polkit-gnome-authentication-agent-1:2815): GLib-CRITICAL **: g_once_init_leave: assertion `initialization_value != 0' failed (null): Warning no default label for /home/hennebry/.gvfs ** Message: applet now removed from the notification area Initializing nautilus-gdu extension Initializing nautilus-open-terminal extension ** Message: applet now embedded in the notification area
To be clear this is the one for CentOS. F14, what I'm running right now, has a different home directory. Is the "... failed" message crash-worthy?
Michael Hennebry wrote:
On Tue, 26 Nov 2013, m.roth@5-cent.us wrote:
Oh, right, *just* as I hit <send>, I realized one more thing: you have checked both ~/.xsession-errors and /var/log/Xorg.0.log, right?
I'd looked at .../Xorg.0.log , but until now, not ~/.xsession-errors :
<snip>
(polkit-gnome-authentication-agent-1:2815): GLib-CRITICAL **: g_once_init_leave: assertion `initialization_value != 0' failed (null): Warning no default label for /home/hennebry/.gvfs
<snip> WAIT ONE MINUTE: "no default label"? Please give us the results of $ getenforce<enter>
mark
On Tue, Nov 26, 2013 at 10:32 AM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
It is getting long in the tooth. (I'm about to replace my m/b of the same
vintage at home.)
Ouch. $200+ before installation, and I'd still have a computer some Linuxes don't like.
Yeah, probably better to get something that is likely to run for 10 more years than to squeeze another year out of something old.
On Mon, 25 Nov 2013, Les Mikesell wrote:
I've seen a machine where it took 3+ days of running memtest86 to catch the error. And then after replacing the RAM, the machine still
Clean bill from memtest86 running overnight.
Like I said, I had one that ran clean for 3 days, then caught the intermittent RAM error that was causing the problem. But, if I hadn't already had a pile of similar RAM to swap in, I would probably have been ahead to trash the old box anyway. It just happened to be the only large tower case I had with an 8 port SATA controller and room for a lot of disks and I like using it for backups.
The memory is the newest part of the computer. After rebooting, it crapped out twice in half an hour. After that, I went back to F14.
Are you saying it never crashes with fedora but it does with centos?
On Tue, 26 Nov 2013, Les Mikesell wrote:
On Tue, Nov 26, 2013 at 10:32 AM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
It is getting long in the tooth. (I'm about to replace my m/b of the same
vintage at home.)
Ouch. $200+ before installation, and I'd still have a computer some Linuxes don't like.
E.g. F16.
Yeah, probably better to get something that is likely to run for 10 more years than to squeeze another year out of something old.
I think that I've seen used laptops for about as much and I would not have to pay for installation.
Are you saying it never crashes with fedora but it does with centos?
F14 crashed once since CentOS started crashing.
Michael Hennebry wrote:
On Tue, 26 Nov 2013, Les Mikesell wrote:
On Tue, Nov 26, 2013 at 10:32 AM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
It is getting long in the tooth. (I'm about to replace my m/b of the same
vintage at home.)
<snip>
Are you saying it never crashes with fedora but it does with centos?
F14 crashed once since CentOS started crashing.
Sounds like CentOS hits something somewhere that FC14 hits more rarely.
mark
On Tue, Nov 26, 2013 at 12:20 PM, m.roth@5-cent.us wrote:
Are you saying it never crashes with fedora but it does with centos?
F14 crashed once since CentOS started crashing.
Sounds like CentOS hits something somewhere that FC14 hits more rarely.
Or just random and too small a sample to draw conclusions.... Back when computer hardware was expensive compared to human time it was fun do debug stuff like that. Now, replacing it with something new would probably save money in the long run just from power usage. Not to mention working 10x faster.
Les Mikesell wrote:
On Tue, Nov 26, 2013 at 12:20 PM, m.roth@5-cent.us wrote:
Are you saying it never crashes with fedora but it does with centos?
F14 crashed once since CentOS started crashing.
Sounds like CentOS hits something somewhere that FC14 hits more rarely.
Or just random and too small a sample to draw conclusions.... Back when computer hardware was expensive compared to human time it was fun do debug stuff like that. Now, replacing it with something new would probably save money in the long run just from power usage. Not to mention working 10x faster.
Depends on the cost of the system, and the budget... and it sounds to me as though the OP is working on his own system, and his budget approaches $1 as a limit.... I understand that all too well (though it's not an issue for me these days).
To the OP: is this a gaming machine, or just what you use? You mentioned $200 before install - I've seen refurbed entire desktops for $300.
mark
On Tue, Nov 26, 2013 at 1:05 PM, m.roth@5-cent.us wrote:
Depends on the cost of the system, and the budget... and it sounds to me as though the OP is working on his own system, and his budget approaches $1 as a limit.... I understand that all too well (though it's not an issue for me these days).
I understand that too, but nothing you can do is going to make it last forever, and power use is a cost if it is on all the time. And I'd recommend something that can virtualize 64-bit OS's (basically anything since a core-2 duo, I think) if you like to try out different things.
To the OP: is this a gaming machine, or just what you use? You mentioned $200 before install - I've seen refurbed entire desktops for $300.
I've had relatively good luck with stuff from tigerdirect.com but you sort of take your chances.
On Tue, 26 Nov 2013, m.roth@5-cent.us wrote:
Depends on the cost of the system, and the budget... and it sounds to me as though the OP is working on his own system, and his budget approaches $1 as a limit.... I understand that all too well (though it's not an issue for me these days).
My current income is zero. I'm burning what little retirement money I have. Some expenses I find more annoying than others. A computer is rather important for a job search these days.
To the OP: is this a gaming machine, or just what you use? You mentioned $200 before install - I've seen refurbed entire desktops for $300.
No. I've never been into gaming. I've got a rather high-end video card because that was the only available AGP card available in a hurry.
The $200 comes from googling D865GBFL price.
$300 desktops? Where?
On Tue, Nov 26, 2013 at 5:58 PM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
The $200 comes from googling D865GBFL price.
$300 desktops? Where?
Search for 'refurbished desktop' on tigerdirect.com.
On Tue, 26 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
On Tue, 26 Nov 2013, m.roth@5-cent.us wrote:
Oh, right, *just* as I hit <send>, I realized one more thing: you have checked both ~/.xsession-errors and /var/log/Xorg.0.log, right?
I'd looked at .../Xorg.0.log , but until now, not ~/.xsession-errors :
<snip> > (polkit-gnome-authentication-agent-1:2815): GLib-CRITICAL **: > g_once_init_leave: assertion `initialization_value != 0' failed > (null): Warning no default label for /home/hennebry/.gvfs <snip> WAIT ONE MINUTE: "no default label"? Please give us the results of $ getenforce<enter>
[hennebry@localhost Desktop]$ getenforce Enforcing
On Tue, 26 Nov 2013, Louis Lagendijk wrote:
You may want to check the bios for some settings. I recently fixed some issues with my server system here at home (gen 1 i7 920) by turning ACPI 2.0 off. I first replaced the power supply but that did improve the situation from disk errors once every day or 2 to once every 5 days. Make sure you have the latest bios as well.
ACPI, but not ACPI 2.0? I'll see what the BIOS claims next time I reboot.
Not at all sure how to update the BIOS. Don't even know whether my machine has a mechanism for updating the BIOS.
Guess what? It died again.
This time I took a look at BIOS stuff. Fan rpms (approximately, two of them changed): Processor region: 3000 Rear: 1500 Front: 0 Perhaps that is my problem. I do have more fans lying around somewhere.
The last even the BIOS logged was from 2008. In case it matters, the BIOS version is BF86510A.86A.0038P08 .
On boot I got a message saying that the CPU was being throttled because it was over the temperature threshold. IIRC the BIOS said that the CPU area temperature was about 80 C.
Michael Hennebry wrote:
Guess what? It died again.
This time I took a look at BIOS stuff. Fan rpms (approximately, two of them changed): Processor region: 3000 Rear: 1500 Front: 0 Perhaps that is my problem. I do have more fans lying around somewhere.
The last even the BIOS logged was from 2008. In case it matters, the BIOS version is BF86510A.86A.0038P08 .
On boot I got a message saying that the CPU was being throttled because it was over the temperature threshold. IIRC the BIOS said that the CPU area temperature was about 80 C.
BING! BING! BING!
You said you'd opened the case - did you clean the heat sink on the CPU? Can you turn the system on with the case cover off, and see if the fan on the CPU is running?
One thing I've never done, or thought of until now, was whether the thermal grease between the CPU and the heat sink had dried out. If it's running hot, that's a possibility, so you might clean that off and put on some new (a buck or so at any computer parts store). Doesn't need much - the force of tightening the heat sink will spread it much farther than you expect it to, and you don't want it coming out the sides.
mark
On Wed, 27 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
Fan rpms (approximately, two of them changed): Processor region: 3000 Rear: 1500 Front: 0
This is ok?
On boot I got a message saying that the CPU was being throttled because it was over the temperature threshold. IIRC the BIOS said that the CPU area temperature was about 80 C.
BING! BING! BING!
You said you'd opened the case - did you clean the heat sink on the CPU?
Not lately I haven't, but it seems that I will soon.
Can you turn the system on with the case cover off, and see if the fan on the CPU is running?
Yes.
One thing I've never done, or thought of until now, was whether the thermal grease between the CPU and the heat sink had dried out. If it's running hot, that's a possibility, so you might clean that off and put on some new (a buck or so at any computer parts store). Doesn't need much - the force of tightening the heat sink will spread it much farther than you expect it to, and you don't want it coming out the sides.
"the force of tightening the heat sink" frightens me silly, but I suppose that would be better than a dead CPU fan. My recollection is that that does not come off.
Michael Hennebry wrote:
On Wed, 27 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
Fan rpms (approximately, two of them changed): Processor region: 3000 Rear: 1500 Front: 0
This is ok?
I have no idea if a) you have a front fan, or b) if it has a separate sensor.
On boot I got a message saying that the CPU was being throttled because it was over the temperature threshold. IIRC the BIOS said that the CPU area temperature was about 80 C.
BING! BING! BING!
You said you'd opened the case - did you clean the heat sink on the CPU?
Not lately I haven't, but it seems that I will soon.
Can you turn the system on with the case cover off, and see if the fan on the CPU is running?
Yes.
One thing I've never done, or thought of until now, was whether the thermal grease between the CPU and the heat sink had dried out. If it's running hot, that's a possibility, so you might clean that off and put on some new (a buck or so at any computer parts store). Doesn't need
much -
the force of tightening the heat sink will spread it much farther than you expect it to, and you don't want it coming out the sides.
"the force of tightening the heat sink" frightens me silly, but I suppose that would be better than a dead CPU fan. My recollection is that that does not come off.
Not to worry. It will probably be a lever that you push down and it catches. I doubt it's like in some servers, where you screw it on... and even in that case, you screw it till you feel it stop turning.
It *really* isn't a Big Deal. These days, nothing's like it was back in the eighties, when taking a system apart was a *lot* of screws, and you could place things the wrong way. For a long time now, they expect people to upgrade or replace parts (cheaper parts, more failures), and if no one else, the zillions of tech support companies leaned on the manufacturers, because they wanted their techs to spend less time per repair.
mark
On Wed, 27 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
On Wed, 27 Nov 2013, m.roth@5-cent.us wrote:
One thing I've never done, or thought of until now, was whether the thermal grease between the CPU and the heat sink had dried out. If it's running hot, that's a possibility, so you might clean that off and put on some new (a buck or so at any computer parts store). Doesn't need
much -
the force of tightening the heat sink will spread it much farther than you expect it to, and you don't want it coming out the sides.
"the force of tightening the heat sink" frightens me silly, but I suppose that would be better than a dead CPU fan. My recollection is that that does not come off.
Not to worry. It will probably be a lever that you push down and it catches. I doubt it's like in some servers, where you screw it on... and even in that case, you screw it till you feel it stop turning.
I found my fans and am about to get some thermal grease and a megohm resistor for static discharge. Sometime today or tomorrow I will likely open the case with fear and trepidation. The sides and top of the case are metal, but painted with an insulator. The front is plastic. The back is metal. I expect I should touch that before opening the case. What about after? Is there something else I should touch before trying to edit its guts?
If thermal grease is the problem, how do I find out and how do I clean off the old stuff? I've read that just adding more is not a good idea. If I add to much thermal paste, what do I do about it?
It *really* isn't a Big Deal. These days, nothing's like it was back in the eighties, when taking a system apart was a *lot* of screws, and you could place things the wrong way. For a long time now, they expect people to upgrade or replace parts (cheaper parts, more failures), and if no one else, the zillions of tech support companies leaned on the manufacturers, because they wanted their techs to spend less time per repair.
Michael Hennebry wrote:
On Wed, 27 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
On Wed, 27 Nov 2013, m.roth@5-cent.us wrote:
One thing I've never done, or thought of until now, was whether the thermal grease between the CPU and the heat sink had dried out. If it's running hot, that's a possibility, so you might clean that off
and put
on some new (a buck or so at any computer parts store). Doesn't need
much - the force of tightening the heat sink will spread it much
farther than
you expect it to, and you don't want it coming out the sides.
"the force of tightening the heat sink" frightens me silly, but I suppose that would be better than a dead CPU fan. My recollection is that that does not come off.
Not to worry. It will probably be a lever that you push down and it catches. I doubt it's like in some servers, where you screw it on... and even in that case, you screw it till you feel it stop turning.
I found my fans and am about to get some thermal grease and a megohm resistor for static discharge. Sometime today or tomorrow I will likely open the case with fear and trepidation. The sides and top of the case are metal, but painted with an insulator. The front is plastic. The back is metal. I expect I should touch that before opening the case. What about after? Is there something else I should touch before trying to edit its guts?
Don't worry. Things are a *lot* less static-sensitive. If you really need grounding, touch a water or gas pipe.
If thermal grease is the problem, how do I find out and how do I clean off the old stuff?
You can start with a paper towel. The FE who was in a month or so ago used an alcohol prep pad.
I've read that just adding more is not a good idea. If I add to much thermal paste, what do I do about it?
I'm still working on "how much". I'd say put a squirt in the middle. Make an
CPU ______ | | | OO | |_____|
Maybe a little more. Don't make a deep puddle - you're just smearing some on. Ever put anti-seize on your spark plugs? <snip> mark
On Tue, 3 Dec 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
I found my fans and am about to get some thermal grease and a megohm resistor for static discharge. Sometime today or tomorrow I will likely open the case with fear and trepidation. The sides and top of the case are metal, but painted with an insulator. The front is plastic. The back is metal. I expect I should touch that before opening the case. What about after? Is there something else I should touch before trying to edit its guts?
Don't worry. Things are a *lot* less static-sensitive. If you really need grounding, touch a water or gas pipe.
I worry. The first time I opened the case, 'twas to install another hard drive. I zapped the video card.
For grounding, a heating duct would be a lot more convenient than a pipe. Would that work?
I've read that just adding more is not a good idea. If I add to much thermal paste, what do I do about it?
I'm still working on "how much". I'd say put a squirt in the middle. Make an
CPU ______ | | | OO | |_____|
Maybe a little more. Don't make a deep puddle - you're just smearing some on. Ever put anti-seize on your spark plugs?
No. I'm a software guy. Even when I did embedded programming, I used the scope a lot less than others.
Michael:
It will be easier and cheaper to by a static grounding strap that you wear on your wrist and connect to a conducting part of the metal cabnet.
http://www.ebay.com/itm/Adjustable-Anti-Static-ESD-Wrist-Strap-Band-w-Ground...
Buy It Now for $0.99
Thomas Dineen
On 12/3/2013 8:27 PM, Michael Hennebry wrote:
On Tue, 3 Dec 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
I found my fans and am about to get some thermal grease and a megohm resistor for static discharge. Sometime today or tomorrow I will likely open the case with fear and trepidation. The sides and top of the case are metal, but painted with an insulator. The front is plastic. The back is metal. I expect I should touch that before opening the case. What about after? Is there something else I should touch before trying to edit its guts?
Don't worry. Things are a *lot* less static-sensitive. If you really need grounding, touch a water or gas pipe.
I worry. The first time I opened the case, 'twas to install another hard drive. I zapped the video card.
For grounding, a heating duct would be a lot more convenient than a pipe. Would that work?
I've read that just adding more is not a good idea. If I add to much thermal paste, what do I do about it?
I'm still working on "how much". I'd say put a squirt in the middle. Make an
CPU ______ | | | OO | |_____|
Maybe a little more. Don't make a deep puddle - you're just smearing some on. Ever put anti-seize on your spark plugs?
No. I'm a software guy. Even when I did embedded programming, I used the scope a lot less than others.
On Wed, 4 Dec 2013, Steve Clark wrote:
Also did you check the Health Status in the BIOS, was the temperature in spec with what the chip spec is. You can run cpuburn (program) and see if running the processor full out causes it to shutdown.
Once upon a time, the CPU area temperature was 80 C, which got a "BING! BING! BING!" from m.roth.
Also, the last time checked I noticed that 5v was 5.263, more than 3% error.
On Tue, 3 Dec 2013, Thomas Dineen wrote:
It will be easier and cheaper to by a static grounding strap that you wear on your wrist and connect to a conducting part of the metal cabnet.
My plan involved a wire-wrap wire "bracelet" and a megohm resistor. Connecting the other end to a bare metal part of the case is sufficient? I don't need an actual ground?
On 12/4/2013 9:42 AM, Michael Hennebry wrote:
Once upon a time, the CPU area temperature was 80 C, which got a "BING! BING! BING!" from m.roth.
Also, the last time checked I noticed that 5v was 5.263, more than 3% error.
I'd only believe that if you double checked it with a known accurate volt meter.
On Wed, 4 Dec 2013, John R Pierce wrote:
On 12/4/2013 9:42 AM, Michael Hennebry wrote:
Once upon a time, the CPU area temperature was 80 C, which got a "BING! BING! BING!" from m.roth.
Also, the last time checked I noticed that 5v was 5.263, more than 3% error.
I'd only believe that if you double checked it with a known accurate volt meter.
From that comment, I infer that were 5.263 the actual voltage,
my computer would be misbehaving much more severely.
Michael Hennebry wrote:
On Wed, 4 Dec 2013, John R Pierce wrote:
On 12/4/2013 9:42 AM, Michael Hennebry wrote:
Once upon a time, the CPU area temperature was 80 C, which got a "BING! BING! BING!" from m.roth.
Also, the last time checked I noticed that 5v was 5.263, more than 3% error.
I'd only believe that if you double checked it with a known accurate volt meter.
From that comment, I infer that were 5.263 the actual voltage,
my computer would be misbehaving much more severely.
Hmmm... that's a though: once you open it up, you might disconnect and measure the output of the power supply with a multimeter. If that's misbehaving, you're talking $40 or so for a new one.
mark
On 12/4/2013 10:22 AM, Michael Hennebry wrote:
On Wed, 4 Dec 2013, John R Pierce wrote:
On 12/4/2013 9:42 AM, Michael Hennebry wrote:
Once upon a time, the CPU area temperature was 80 C, which got a "BING! BING! BING!" from m.roth.
Also, the last time checked I noticed that 5v was 5.263, more than 3% error.
I'd only believe that if you double checked it with a known accurate volt meter.
From that comment, I infer that were 5.263 the actual voltage,
my computer would be misbehaving much more severely.
almost nothing in the system actually runs directly on 5V anymore. the CPU runs on some fraction of a volt at stupid high current, generated by DC-DC regulators which use the 12V supply as their input. Ditto, the ram runs on like 1.8V nowdays, same thing, regulated off 12V. most peripheral logic is 3V or less nowdays.
disk drives use 5V (and 3.5" drives use 12V) but they also ahve their own internal voltage converters and regulators.
On 12/05/2013 06:42 AM, Michael Hennebry wrote:
My plan involved a wire-wrap wire "bracelet" and a megohm resistor. Connecting the other end to a bare metal part of the case is sufficient? I don't need an actual ground?
No you need an actual ground. If the case is left plugged in this would be sufficient but is probably a bad idea for other (obvious) reasons. Find something else to attach the strap (or whatever) to and make sure it's grounded.
Peter
On 12/4/2013 11:25 AM, Peter wrote:
On 12/05/2013 06:42 AM, Michael Hennebry wrote:
My plan involved a wire-wrap wire "bracelet" and a megohm resistor. Connecting the other end to a bare metal part of the case is sufficient? I don't need an actual ground?
No you need an actual ground. If the case is left plugged in this would be sufficient but is probably a bad idea for other (obvious) reasons. Find something else to attach the strap (or whatever) to and make sure it's grounded.
the case has to be grounded to the same ground or it won't do much good.
I've never had any issues with static as long as I touch the metal chassis first before handling circuitry, but I also live on the coast where we don't need air conditioning, and hte humidity is moderate.
On Wed, 4 Dec 2013, John R Pierce wrote:
On 12/4/2013 11:25 AM, Peter wrote:
On 12/05/2013 06:42 AM, Michael Hennebry wrote:
My plan involved a wire-wrap wire "bracelet" and a megohm resistor. Connecting the other end to a bare metal part of the case is sufficient? I don't need an actual ground?
No you need an actual ground. If the case is left plugged in this would be sufficient but is probably a bad idea for other (obvious) reasons. Find something else to attach the strap (or whatever) to and make sure it's grounded.
the case has to be grounded to the same ground or it won't do much good.
The new plan is to connect myself to the case through a megohm resistor and the case to a heating vent through another megohm resistor.
My money's still on bulging and 50% or so under (if not ruptured and 99% below) capacitance electrolytics. If not on the motherboard itself, then in the power supply.
On 12/03/2013 10:16 PM, Michael Hennebry wrote:
I found my fans and am about to get some thermal
Make sure you make a note in which direction all the fans in the PC are blowing. Usually there is an arrow on them which tells you which way they blow but you can also feel it by holding your hand in front of them. The replacement fans need to blow in the same direction so think about that when putting them in.
grease and a megohm resistor for static discharge. Sometime today or tomorrow I will likely open the case with fear and trepidation.
Don't worry. It's nothing like it was in the eighties when stuff fell apart by merely looking at it (except for the IBM keyboards).
The sides and top of the case are metal, but painted with an insulator. The front is plastic. The back is metal. I expect I should touch that before opening the case.
As far as I know touching something that's properly grounded should do it. Maybe something like gas/water/heating pipes (unpainted bare metal). Stating the obvious but please do disconnect the power cord before doing anything and wait a minute. If the power supply itself has an on/off switch (usually at the back) then leave the switch on and disconnect the power cord. If it also has a light you can see it go dark. Even after the power supply has been disconnected it can still have a charge so don't go poking any metal objects in there unless you want smoke coming out of your ears.
What about after? Is there something else I should touch before trying to edit its guts?
Don't think so but refrain from touching the actual chips. And do it near a lamp with a lot of light.
If thermal grease is the problem, how do I find out and how do I clean off the old stuff?
There are a lot of instructions here: http://www.arcticsilver.com/intel_application_method.html#
I've read that just adding more is not a good idea.
Correct. You only need a really small amount of it. It's only needed to fill any air pockets (=lot's of heat getting trapped) with thermal paste between the cpu and the heatsink so the heat is guided away through the heatsink instead of getting stuck and frying your cpu.
Clean both the heatsink and the cpu so the old stuff is removed. Only then you apply thermal paste only on the cpu (not on the heatsink). Read the instructions at the link I gave you earlier.
If I add to much thermal paste, what do I do about it?
Remove it with a credit card or something non-metallic similar. Read through the instructions at the link I gave you earlier.
I usually remove dust with a vacuumcleaner where I can without touching anything in the PC. If you want to do it the fancy way get a can of compressed air and blow the dust straight out.
Good luck.
Regards, Patrick
On Tue, 03 Dec 2013 23:25:37 +0100 Patrick Lists wrote:
If you want to do it the fancy way get a can of compressed air and blow the dust straight out.
Bad plan. Canned air generally blows dust into areas that you don't want it.
I use canned air to blow dust off of things like monitor surfaces, but not inside of the computer itself. A small vacuum works well for that purpose.
On Tue, 3 Dec 2013, Patrick Lists wrote:
There are a lot of instructions here: http://www.arcticsilver.com/intel_application_method.html#
They mention "high-purity isopropyl alcohol". The highest purity I've ever seen is 70%.
Also, said instructions do recommend spreading some thermal grease manually (actually, redit-cardily). One wipes off as much as one can before adding the blob to be spread by the heat spreader.
On 12/3/2013 9:09 PM, Michael Hennebry wrote:
There are a lot of instructions here:
They mention "high-purity isopropyl alcohol". The highest purity I've ever seen is 70%.
isopropyl commonly comes in 70% ('rubbing alcohol') and 90% (used for wiping injection sites and such).
Also, said instructions do recommend spreading some thermal grease manually (actually, redit-cardily). One wipes off as much as one can before adding the blob to be spread by the heat spreader.
the application method varies with the CPU type on that page. some suggest just a dot, others a line, and yet others doing the manual spread.
On 12/3/2013 22:09, Michael Hennebry wrote:
On Tue, 3 Dec 2013, Patrick Lists wrote:
They mention "high-purity isopropyl alcohol". The highest purity I've ever seen is 70%.
The vast majority of the impurity in commercial grade alcohol is water. Capillary action sucks the fluid into all kinds of areas you might not expect -- like under chips -- and it stays there longer than you might guess because restricted airflow goes hand in hand with capillary action. Also, water evaporates far less readily than alcohol, increasing drying time.
The rest of the impurity is dissolved solids, which gets left behind when the liquid evaporates. It also makes the fluid conductive. (Pure alcohol and pure water are *not* conductive.) Conductive liquid is obviously bad for computers, especially if it's still present when you apply power. See previous paragraph. :)
Bottom line: 70% is too impure for this task.
I hesitate to use even 90% for this. The last time I used 90% isopropyl on a PCB, it left behind a white haze that I had to scrub off with a dry toothbrush.
I bought a box (!) of 99% isopropyl years ago: http://goo.gl/7DYP8Y
It's fairly expensive to ship, but even so, it comes out cheaper than the alternatives you're likely to have locally. Figure $0.30 - $0.40 per ounce, all told. You'll probably be set for life. (Tip: Add a Menda bottle for your work table to your order, so you can keep the box safe, like in a shed or garage.)
Radio Shack used to sell tape head cleaner for $1 an ounce. I'd guess it's no longer available because there's not as much call for tape head cleaning products these days.
While looking for up-to-date info on Radio Shack's web site, I came across this relevant item: http://goo.gl/nLzub7 At $11 for a couple of tiny bottles, both of which you have to use together, it's another 2-3x more expensive than tape head cleaner. Plus, if you look into the MSDS, the first part is acidic, so you must need the second pass to neutralize what's left behind on the first pass. Sounds like a bad plan to me.
Everclear 190 proof should also work for this. At $20 per fifth, it comes out under $1/oz, so cheaper than the RS fluids, but still more expensive than the box o' isopropyl. Some households will find it a more widely useful commodity, so there's that. :)
For what it's worth, you can get even purer isopropyl alcohol intended for lab use. Prices I found online ranged from about $60-100 per liter, or $2-3/oz, shipped. 100% is possible if you synthesize it, at even higher cost.
On Wed, 4 Dec 2013, Warren Young wrote:
On 12/3/2013 22:09, Michael Hennebry wrote:
They mention "high-purity isopropyl alcohol". The highest purity I've ever seen is 70%.
Bottom line: 70% is too impure for this task.
I hesitate to use even 90% for this. The last time I used 90% isopropyl
While looking for up-to-date info on Radio Shack's web site, I came across this relevant item: http://goo.gl/nLzub7 At $11 for a couple of tiny bottles, both of which you have to use together, it's another 2-3x more expensive than tape head cleaner. Plus, if you look into the MSDS, the first part is acidic, so you must need the second pass to neutralize what's left behind on the first pass. Sounds like a bad plan to me.
For me, the bottom line is how much it will cost to clean *one* CPU/heat spreader combination. It looks like the answer is going to be a bottle I saw at Radio Shack for $11.
Everclear 190 proof should also work for this. At $20 per fifth, it comes out under $1/oz, so cheaper than the RS fluids, but still more expensive than the box o' isopropyl. Some households will find it a more widely useful commodity, so there's that. :)
I didn't use everclear when I did drink.
For what it's worth, you can get even purer isopropyl alcohol intended for lab use. Prices I found online ranged from about $60-100 per liter, or $2-3/oz, shipped. 100% is possible if you synthesize it, at even higher cost.
I'm not sure what you mean by synthesize in this context. Alternating freeze- and evaporative-distillation can get you anything less than 100%. That said, 'tain't necessarily economical. At some point, I suspect getting rid of the water through electrolysis or some other chemistry might be better.
On Dec 4, 2013, at 1:22 PM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
On Wed, 4 Dec 2013, Warren Young wrote:
At $11 for a couple of tiny bottles, both of which you have to use together, it's another 2-3x more expensive than tape head cleaner.
For me, the bottom line is how much it will cost to clean *one* CPU/heat spreader combination.
Because I always have a Menda bottle full of high-purity alcohol on the electronics bench, I find many uses for it.
If I had to drive to Radio Shack every ounce or so to get a refill, I would doubtless not bother for most jobs.
100% is possible if you synthesize it
I'm not sure what you mean by synthesize in this context.
I mean assemble the molecules from bulk quantities of their constituent elements. :) Chemical engineering.
As my measurement data in the post I just sent hints, there are serious practical problems -- aside from the direct economic ones -- that restrict the utility of such high-purity alcohols. There's probably no point using it outside a cleanroom. Dust on the glassware will wreck the purity rating otherwise.
On 12/4/2013 8:00 PM, Warren Young wrote:
As my measurement data in the post I just sent hints, there are serious practical problems -- aside from the direct economic ones -- that restrict the utility of such high-purity alcohols. There's probably no point using it outside a cleanroom. Dust on the glassware will wreck the purity rating otherwise.
alcohol much higher than ~ 90% is hydroscopic and will absorb humidity from the air until it reaches the azeotriopic place where its happy.
seriously, its a heat spreader, youre gonna be spreading silicone paste on. any old alcohol will work fine, don't use /too/ much, just dampen the wipes with it, repeat with clean wipes til the surface is clean, air dry in a few seconds, then apply fresh arctic silver v or whatever. frankly, for servers, I prefer using the plain white classic thermal silicone stuff, its less conductive, and servers usually have plenty of air and cooling, you're not overclocking them.
On Wed, 4 Dec 2013, Warren Young wrote:
On Dec 4, 2013, at 1:22 PM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
On Wed, 4 Dec 2013, Warren Young wrote:
At $11 for a couple of tiny bottles, both of which you have to use together, it's another 2-3x more expensive than tape head cleaner.
For me, the bottom line is how much it will cost to clean *one* CPU/heat spreader combination.
Because I always have a Menda bottle full of high-purity alcohol on the electronics bench, I find many uses for it.
If I had to drive to Radio Shack every ounce or so to get a refill, I would doubtless not bother for most jobs.
I don't have an electronics bench. For me, this is a Very Special Occasion, one that I hope notto repeat for a long time.
100% is possible if you synthesize it
I'm not sure what you mean by synthesize in this context.
I mean assemble the molecules from bulk quantities of their constituent elements. :) Chemical engineering.
That is what I thought you meant. I'm not sure that is even possible. Judging from Wikipedia, no one makes it that way.
As my measurement data in the post I just sent hints, there are serious practical problems -- aside from the direct economic ones -- that restrict the utility of such high-purity alcohols. There's probably no point using it outside a cleanroom. Dust on the glassware will wreck the purity rating otherwise.
I wasn't planning to look for 99.9% pure isopropyl alcohol.
On 2013-12-05 2:51 PM, Michael Hennebry wrote:
I wasn't planning to look for 99.9% pure isopropyl alcohol.
I buy 99% isopropyl at Meijer Thrifty Acres (grocery store chain here in the midwest). http://i41.photobucket.com/albums/e273/Darr247/Electronics/198-ProofIsopropy... I think that bottle was about $1.50 (price tags are no longer required by law, so I'm not certain).
I use Dow-Corning 340 as heat-sink paste, by the way... compared to Arctic Silver it's about 1/5th the cost and ~99.5% of its heat-transfer performance (and I've never seen it dry out like I have seen Arctic Silver do after a few years). e.g. 1oz - http://www.amazon.com/dp/B00D0HVYHI 2oz - http://www.amazon.com/dp/B00CB2K3LG There should be enough in that 2oz jar to do/redo dozens of CPUs (or any other component with a heat sink). Unless you do a couple every week, the 140g tube (http://www.amazon.com/dp/B00CAVTGNE) would likely reach its expiration date long before it's half gone.
On 12/4/2013 10:44 AM, Warren Young wrote:
Bottom line: 70% is too impure for this task.
Huh?!? I've cleaned numerous CPU-heatsink surfaces with 70% isopropyl, never had any problem.... heck, I clean optical lenses with it, in the form of those eyeglass wipes you buy by the crate at Costco.
in tests we did 30 something years ago, the 90% medical stuff left a white haze, while the 70% regular rubbing stuff didn't. we were using it for cleaning the heads and tape paths of 9 track computer tape drives. sure, the analytic laboratory grade was clean, but also stupid expensive, as we were going through a quart a week cleaning a dozen high speed tape drives 3 shifts a day 6 days a week (they were in constant use loading tapes shipped to us from vendors, so got rather dirty rather fast).
On Dec 4, 2013, at 1:49 PM, John R Pierce pierce@hogranch.com wrote:
On 12/4/2013 10:44 AM, Warren Young wrote:
Bottom line: 70% is too impure for this task.
Huh?!? I've cleaned numerous CPU-heatsink surfaces with 70% isopropyl, never had any problem….
I'm not saying it's impossible. An expert can ameliorate many risks through skill. That is no reason to recommend a risky process.
I listed the theoretical risks in my previous post. Now for some experimental data.
I filled a plastic bowl to a depth of 1 inch with the contents of a 21-year-old bottle of 70% rubbing alcohol from my medicine cabinet. (Ah, 1992…it was a good year.) I then inserted my bench DMM's probes into the bowl, spacing them approximately 1 inch apart. I read 377 kΩ.
I sanity checked this measurement by moving the probes closer to each other, then farther apart, and observed that the resistance changed as expected.
I poured the rubbing alcohol back into the bottle, dried the bowl with a towel, poured the contents of my electronics bench's Menda bottle -- nominally 99+% pure isopropyl alcohol -- into the bowl, then tested again. This time I read over 2 MΩ.
I was expecting a higher value, or even an overload indication. I decided to refill the Menda bottle straight from the cubitainer in order to rule out impurities in the bowl, and also the natural concentration of impurities due to evaporation from the Menda bottle. This time I read over 8 MΩ.
You questioned someone else's DMM in another post, so I will pre-defend mine. It's a general purpose bench meter, not a dedicated insulation tester, but it will go up to 1 GΩ, and the company that made the meter isn't the sort that publishes bogus specs. It's been a while since it was calibrated, but for 3 digit readings, I'm confident enough in its quality of design and manufacture to trust those readings anyway.
I clean optical lenses with it, in the form of those eyeglass wipes you buy by the crate at Costco.
I use my tee shirt. :)
On 12/4/2013 7:50 PM, Warren Young wrote:
You questioned someone else's DMM in another post,
actually, the other post didn't say how he measured the voltage, I was assuming via the motherboard monitoring circuits ('lmsensors' or whatever), which are notoriously inaccurate ...
On Wed, 4 Dec 2013, John R Pierce wrote:
On 12/4/2013 7:50 PM, Warren Young wrote:
You questioned someone else's DMM in another post,
actually, the other post didn't say how he measured the voltage, I was assuming via the motherboard monitoring circuits ('lmsensors' or whatever), which are notoriously inaccurate ...
I'd thought 'twas obvious that I'd used the only method available without opening the case: BIOS.
On Tue, 2013-12-03 at 15:16 -0600, Michael Hennebry wrote:
On Wed, 27 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
On Wed, 27 Nov 2013, m.roth@5-cent.us wrote:
One thing I've never done, or thought of until now, was whether the thermal grease between the CPU and the heat sink had dried out. If it's running hot, that's a possibility, so you might clean that off and put on some new (a buck or so at any computer parts store). Doesn't need
much -
the force of tightening the heat sink will spread it much farther than you expect it to, and you don't want it coming out the sides.
"the force of tightening the heat sink" frightens me silly, but I suppose that would be better than a dead CPU fan. My recollection is that that does not come off.
Not to worry. It will probably be a lever that you push down and it catches. I doubt it's like in some servers, where you screw it on... and even in that case, you screw it till you feel it stop turning.
In any case: check the fans first. There may also be a lot of dirt on the heat sink. These are much more likely culprits than the thermal paste.
If thermal grease is the problem, how do I find out and how do I clean off the old stuff? I've read that just adding more is not a good idea. If I add to much thermal paste, what do I do about it?
If everything else fails you can try to replace the old thermal paste. I It seems unlikely that this is the cause of your problems. I have seen quite a number of issues with fans and dirty heat sinks. None that I can remember of bad thermal paste. I may have seen one case where there was none at all (assembly mistake). Remember that the paste is only to be used to fill up the really small unevenness between heat sink and heat spreader on the processor. Unless you are overclocking I would not expect much difference from a new thermal paste
As mark already pointed out a little alcohol is very helpful to remove old paste. Use a lint free cloth to remove it. Then just put a little new paste on in the middle of the processor and use a credit card to spread it out as thin as possible. The credit card is flexible enough to follow the surface accurately. And don't let the paste spread out from the sides of the heat spreader.
Louis
On Tue, Dec 3, 2013 at 5:59 PM, Louis Lagendijk louis@fazant.net wrote:
On Tue, 2013-12-03 at 15:16 -0600, Michael Hennebry wrote:
On Wed, 27 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
On Wed, 27 Nov 2013, m.roth@5-cent.us wrote:
One thing I've never done, or thought of until now, was whether the thermal grease between the CPU and the heat sink had dried out. If
it's
running hot, that's a possibility, so you might clean that off and
put
on some new (a buck or so at any computer parts store). Doesn't need
much -
the force of tightening the heat sink will spread it much farther
than
you expect it to, and you don't want it coming out the sides.
"the force of tightening the heat sink" frightens me silly, but I suppose that would be better than a dead CPU fan. My recollection is that that does not come off.
Not to worry. It will probably be a lever that you push down and it catches. I doubt it's like in some servers, where you screw it on...
and
even in that case, you screw it till you feel it stop turning.
In any case: check the fans first. There may also be a lot of dirt on the heat sink. These are much more likely culprits than the thermal paste.
If thermal grease is the problem, how do I find out and how do I clean off the old stuff? I've read that just adding more is not a good idea. If I add to much thermal paste, what do I do about it?
A drop about the size of a pea.
Use the heat sink to spread it out that way there won't be any air bubbles. Just put the heat sink on top, then use the clips (or whatever fasteners you took off) to secure the heat sink (and it spreads the thermal paste too).
*If* you put too much, you'll probably have some gushing out the sides. I've not had problems when I put a dot/dab the size of a pea.
If everything else fails you can try to replace the old thermal paste. I It seems unlikely that this is the cause of your problems. I have seen quite a number of issues with fans and dirty heat sinks. None that I can remember of bad thermal paste. I may have seen one case where there was none at all (assembly mistake). Remember that the paste is only to be used to fill up the really small unevenness between heat sink and heat spreader on the processor. Unless you are overclocking I would not expect much difference from a new thermal paste
As mark already pointed out a little alcohol is very helpful to remove old paste. Use a lint free cloth to remove it. Then just put a little new paste on in the middle of the processor and use a credit card to spread it out as thin as possible. The credit card is flexible enough to follow the surface accurately. And don't let the paste spread out from the sides of the heat spreader.
Don't spread the thermal paste with anything - there's no need. Use the heat sink! No air bubbles/pockets!
Louis
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
On 11/26/2013 3:58 PM, Michael Hennebry wrote:
The $200 comes from googling D865GBFL price.
that board is expensive because its ancient, there's likely very little NOS (new old stock) left in the channel.
the 865G chipset was new in May 2003. An Atom based netbook has more horsepower. so does an android phone.
re: $300 desktops, yes you can find them, usually last years consumer big-box store crap-quality, on clearance sale... frankly, those kinds of systems really aren't very good quality, although if/when they work, they'll do the job
On Tue, Nov 26, 2013 at 6:58 PM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
On Tue, 26 Nov 2013, m.roth@5-cent.us wrote:
Depends on the cost of the system, and the budget... and it sounds to me as though the OP is working on his own system, and his budget approaches $1 as a limit.... I understand that all too well (though it's not an issue for me these days).
My current income is zero. I'm burning what little retirement money I have. Some expenses I find more annoying than others. A computer is rather important for a job search these days.
To the OP: is this a gaming machine, or just what you use? You mentioned $200 before install - I've seen refurbed entire desktops for $300.
No. I've never been into gaming. I've got a rather high-end video card because that was the only available AGP card available in a hurry.
The $200 comes from googling D865GBFL price.
$300 desktops? Where?
http://www.newegg.com/Product/Product.aspx?Item=N82E16883155832. They also have a few older (core2 duos) desktop for $200 or less.
-- Michael hennebry@web.cs.ndsu.NoDak.edu "On Monday, I'm gonna have to tell my kindergarten class, whom I teach not to run with scissors, that my fiance ran me through with a broadsword." -- Lily _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
On 11/26/2013 3:58 PM, Michael Hennebry wrote:
$300 desktops? Where?
how about under $100 for a complete desktop?
http://www.newegg.com/Special/ShellShocker.aspx?nm_mc=EMC-SD112013&cm_mm...
you'll need to subscribe to newegg's shellshocker list before you can order it, and this deal is only 'visible' and valid from 3-6pm PST tomorrow (Thanksgiving Day), its a refurb Dell OptiPlex GX740 stripped model, miditower with an AMD Athlon 64 X2 3800+ dual core 2Ghz, 2GB ram, 80GB HD, keyboard, mouse. For $95 (after a $10 mail-in-rebate). I'd want to add at least 4GB more memory (it will take max 4 x 2GB == 8GB DDR2 dimms... I suspect it comes with 2 x 1GB), and my existing SATA disks (it has 4 SATA ports).
Hey there,
He has a 3.2 Ghz CPU which is much more then this 3800 but the only thing that the 3800+ is good is that it has 2 cores but still same low share ram ( I had one of these in the past).
Eliezer
On 28/11/13 12:29, John R Pierce wrote:
On 11/26/2013 3:58 PM, Michael Hennebry wrote:
$300 desktops? Where?
how about under $100 for a complete desktop?
http://www.newegg.com/Special/ShellShocker.aspx?nm_mc=EMC-SD112013&cm_mm...
you'll need to subscribe to newegg's shellshocker list before you can order it, and this deal is only 'visible' and valid from 3-6pm PST tomorrow (Thanksgiving Day), its a refurb Dell OptiPlex GX740 stripped model, miditower with an AMD Athlon 64 X2 3800+ dual core 2Ghz, 2GB ram, 80GB HD, keyboard, mouse. For $95 (after a $10 mail-in-rebate). I'd want to add at least 4GB more memory (it will take max 4 x 2GB == 8GB DDR2 dimms... I suspect it comes with 2 x 1GB), and my existing SATA disks (it has 4 SATA ports).
On Wed, 4 Dec 2013, Eliezer Croitoru wrote:
He has a 3.2 Ghz CPU which is much more then this 3800 but the only thing that the 3800+ is good is that it has 2 cores but still same low share ram ( I had one of these in the past).
I'm considering the beast listed here, especially if my current beast dies: http://www.newegg.com/Product/Product.aspx?Item=N82E16883250296&cm_sp=Da... I note that it has an Intel Core 2 Duo and 2G of RAM. It has those in common with a lot of cheap PC's I've seen. Are there gotchas here that I should know about? I'm already aware that Core 2 is on Intel's discontiued list. My current beast has Pentium 4 with 4G of RAM.
On Thu, Dec 5, 2013 at 2:02 PM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
He has a 3.2 Ghz CPU which is much more then this 3800 but the only thing that the 3800+ is good is that it has 2 cores but still same low share ram ( I had one of these in the past).
I'm considering the beast listed here, especially if my current beast dies: http://www.newegg.com/Product/Product.aspx?Item=N82E16883250296&cm_sp=Da... I note that it has an Intel Core 2 Duo and 2G of RAM. It has those in common with a lot of cheap PC's I've seen. Are there gotchas here that I should know about? I'm already aware that Core 2 is on Intel's discontiued list. My current beast has Pentium 4 with 4G of RAM.
Depending on what you are doing, you might miss the RAM. You should be able to enable the VTx capability on that if you want to run KVM but the option is hidden somewhere in the bios.
Les Mikesell wrote:
On Thu, Dec 5, 2013 at 2:02 PM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
He has a 3.2 Ghz CPU which is much more then this 3800 but the only thing that the 3800+ is good is that it has 2 cores but still same low share ram ( I had one of these in the past).
I'm considering the beast listed here, especially if my current beast dies: http://www.newegg.com/Product/Product.aspx?Item=N82E16883250296&cm_sp=Da... I note that it has an Intel Core 2 Duo and 2G of RAM. It has those in common with a lot of cheap PC's I've seen. Are there gotchas here that I should know about? I'm already aware that Core 2 is on Intel's discontiued list. My current beast has Pentium 4 with 4G of RAM.
Depending on what you are doing, you might miss the RAM. You should be able to enable the VTx capability on that if you want to run KVM but the option is hidden somewhere in the bios.
That's a nice price, but you really do want at least 4G of RAM. You might see if you can add more, at least two gig - bring it up to 4GB.
mark
What Refurbished means?
The hardware by itself looks nice but it might be a noisy machine. HP support only windows Vista for this machine and I do not know what bios and CHIPS it was built upon yet.
If it has the parts that this review claims: http://reviews.cnet.com/desktops/hp-compaq-business-dc7800/4507-3118_7-32598...
Then this is the official compatibility the chipset has: http://www.intel.com/p/en_US/support/category/graphics/q35/cmptbl
I have not used yet a q35 based machine but it seems like kvm is using\planing this chipset design as an emulated layer in kvm: http://www.linux-kvm.org/wiki/images/0/06/2012-forum-Q35.pdf
So the basic assumption is that it was used by a client and then was replaced by newer desktop to prevent something or just to move forward.
Compared to the +3800X2 I think E6750 requires more Power but it has VT-x support.
Due to "2.33" I assume it's not the E6750 but maybe E6550. And as long you do not expect it to lift your desktop to the air it should be a good machine.
Compared to INTEL ATOM it is rated for 65w which most ATOM are about 15w.
Basic EMAIL(not 40k emails) and basic browsing(not too much concurrently open tabs) should run simultaneously by default. If you can buy the 4GB as a package it will give you more air to breath while comparing it to the old machine.
The HDD state is irrelevant as I understand.
The basic issue with this machine is that in my part of the world I cannot get replacement parts for it.
If you can try to ask in nearby small stores what MB they do have and what parts are the basic ones today you will know what to expect.
Eliezer
On 05/12/13 22:02, Michael Hennebry wrote:
I'm considering the beast listed here, especially if my current beast dies: http://www.newegg.com/Product/Product.aspx?Item=N82E16883250296&cm_sp=Da... I note that it has an Intel Core 2 Duo and 2G of RAM. It has those in common with a lot of cheap PC's I've seen. Are there gotchas here that I should know about? I'm already aware that Core 2 is on Intel's discontiued list. My current beast has Pentium 4 with 4G of RAM.
On Sat, 7 Dec 2013, Eliezer Croitoru wrote:
What Refurbished means?
newegg: "Refurbished" products have been tested to ensure compliance with original manufacturer specifications, and MAY include a limited manufacturer warranty - see the item's product page for details."
The hardware by itself looks nice but it might be a noisy machine. HP support only windows Vista for this machine and I do not know what bios and CHIPS it was built upon yet.
If it has the parts that this review claims: http://reviews.cnet.com/desktops/hp-compaq-business-dc7800/4507-3118_7-32598...
I noticed the TPM 1.2 . Am I going have to start dealing with the "trusted" computing crap when I get another computer?
Also, I just noticed that the seller's answer to a question suggests that the power supply is borderline.
Due to "2.33" I assume it's not the E6750 but maybe E6550. And as long you do not expect it to lift your desktop to the air it should be a good machine.
So no gotchas with Core 2 Duo and 2G of RAM.
Compared to INTEL ATOM it is rated for 65w which most ATOM are about 15w.
Basic EMAIL(not 40k emails) and basic browsing(not too much concurrently open tabs) should run simultaneously by default.
Development work.
If you can buy the 4GB as a package it will give you more air to breath while comparing it to the old machine.
On 05/12/13 22:02, Michael Hennebry wrote:
I'm considering the beast listed here, especially if my current beast dies: http://www.newegg.com/Product/Product.aspx?Item=N82E16883250296&cm_sp=Da... I note that it has an Intel Core 2 Duo and 2G of RAM. It has those in common with a lot of cheap PC's I've seen. Are there gotchas here that I should know about? I'm already aware that Core 2 is on Intel's discontiued list. My current beast has Pentium 4 with 4G of RAM.
Hey Michael,
The TPM can be an issue but once you install the OS(LINUX) on DISK(maybe on another machine) it should fly by default. What is the meaning of "Trusted" by HP\COMPAQ? I do not know yet.
From what I understand a Refurbished means "Used and was used in a company\office the last time before being tested in lab".
gotchas with RAM ? like what? If it's a RAM that can take your workload for more then a week it is basically fine. Only when you see weird stuff happening on the desktop you do understand that there is something wrong. You can do the same like in servers that each and every boot do a full slow memory tests with the only difference: it do not have any ECC check at all.
Do you run compiling jobs on this machine and\or plain Coding ? others?
Eliezer
On 07/12/13 09:09, Michael Hennebry wrote:
On Sat, 7 Dec 2013, Eliezer Croitoru wrote:
What Refurbished means?
newegg: "Refurbished" products have been tested to ensure compliance with original manufacturer specifications, and MAY include a limited manufacturer warranty - see the item's product page for details."
The hardware by itself looks nice but it might be a noisy machine. HP support only windows Vista for this machine and I do not know what bios and CHIPS it was built upon yet.
If it has the parts that this review claims: http://reviews.cnet.com/desktops/hp-compaq-business-dc7800/4507-3118_7-32598...
I noticed the TPM 1.2 . Am I going have to start dealing with the "trusted" computing crap when I get another computer?
Also, I just noticed that the seller's answer to a question suggests that the power supply is borderline.
Due to "2.33" I assume it's not the E6750 but maybe E6550. And as long you do not expect it to lift your desktop to the air it should be a good machine.
So no gotchas with Core 2 Duo and 2G of RAM.
Compared to INTEL ATOM it is rated for 65w which most ATOM are about 15w.
Basic EMAIL(not 40k emails) and basic browsing(not too much concurrently open tabs) should run simultaneously by default.
Development work.
If you can buy the 4GB as a package it will give you more air to breath while comparing it to the old machine.
On 05/12/13 22:02, Michael Hennebry wrote:
I'm considering the beast listed here, especially if my current beast dies: http://www.newegg.com/Product/Product.aspx?Item=N82E16883250296&cm_sp=Da... I note that it has an Intel Core 2 Duo and 2G of RAM. It has those in common with a lot of cheap PC's I've seen. Are there gotchas here that I should know about? I'm already aware that Core 2 is on Intel's discontiued list. My current beast has Pentium 4 with 4G of RAM.
On 12/6/2013 11:47 PM, Eliezer Croitoru wrote:
The TPM can be an issue but once you install the OS(LINUX) on DISK(maybe on another machine) it should fly by default. What is the meaning of "Trusted" by HP\COMPAQ? I do not know yet.
From what I understand a Refurbished means "Used and was used in a company\office the last time before being tested in lab".
refurbs can also be open box mechandise. retail product returns. at dell outlet, a lot of the systems they sell are production overruns and order cancellations. XYZ Corp ordered 30 of a specific configuration Latitude E7440, then cut back to 20, so Dell has 10 extra. I've bought stuff from Dell Outlet, both the business version and the home version, its been fine. you have to check it a couple times a day for a few weeks if you want a specific configuration (say, a E6420 14", with the FHD screen, the backlit keyboard, a good CPU, and extended battery).
On 12/6/2013 11:09 PM, Michael Hennebry wrote:
I noticed the TPM 1.2 . Am I going have to start dealing with the "trusted" computing crap when I get another computer?
TPM is totally optional to use.
it can be used so your system can establish a chain of trust with a server or network or whatever, but there's absolutely no requirement to use it. it also can be used as a trust store for secureboot and full disk encryption... If the OS has been secured this way, and you reset the TPM via the BIOS, you will have to reformat the disks to use them.
we looked at TPM for authenticating unattended clients making ssl connections, but getting all that working just so seemed a little too sketchy so we abandoned the idea.
Michael Hennebry wrote:
On Sun, 24 Nov 2013, John R Pierce wrote:
On 11/24/2013 9:45 PM, Michael Hennebry wrote:
CentOS 6.4 died on me again.
only time that has EVER happened to me, on dozens and dozens of systems, has been when there's been a serious hardware problem.
I really do not know whether to hope you are correct. On one hand a new computer would be expensive. On the other, if it's something else, my diagnostic skills are clearly not up to the task.
What kind of m/b is this, and what are you running on it? We've had that happen occasionally on the 48 and 64 core Penguins, which are all Supermicro... and we've had to send a number of them back for repair under warranty, including some several times. Supermicro seems to be a bit light on the concept of q/a q/c....
mark
On Mon, 25 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
On Sun, 24 Nov 2013, John R Pierce wrote:
On 11/24/2013 9:45 PM, Michael Hennebry wrote:
CentOS 6.4 died on me again.
only time that has EVER happened to me, on dozens and dozens of systems, has been when there's been a serious hardware problem.
I really do not know whether to hope you are correct. On one hand a new computer would be expensive. On the other, if it's something else, my diagnostic skills are clearly not up to the task.
What kind of m/b is this, and what are you running on it? We've had that happen occasionally on the 48 and 64 core Penguins, which are all
The computer is a DakTEch Freedom 4 P4 DDR System. The system board is a D865GBFL w/LAN,audio & video Processor Intel Pentium 4 3.2 Ghz 800FSB I got it in 2006. I switched to CentOS because Fedora will not install on it any more. Fedora 14 is the last I was able to install. Installation has almost always been a tremendous hassle for me, so I've usually not gone with the latest and greatest until my current nears EOL. I've read that a kernel bug is the reason that I could not install F16. Supposedly it had been fixed by F17, but no go.
On 11/25/2013 8:58 AM, Michael Hennebry wrote:
The computer is a DakTEch Freedom 4 P4 DDR System. The system board is a D865GBFL w/LAN,audio & video Processor Intel Pentium 4 3.2 Ghz 800FSB I got it in 2006.
its past the age where components are starting to fail. I'd suspect the capacitors on the mainboard, power supply, the DC DC regulators that supply the CPU power, etc etc. I've found over the years most electronics stuff has a half life of 5 years (half of it fails in 5 years).
Michael Hennebry wrote:
On Mon, 25 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
On Sun, 24 Nov 2013, John R Pierce wrote:
On 11/24/2013 9:45 PM, Michael Hennebry wrote:
CentOS 6.4 died on me again.
only time that has EVER happened to me, on dozens and dozens of systems, has been when there's been a serious hardware problem.
I really do not know whether to hope you are correct. On one hand a new computer would be expensive. On the other, if it's something else, my diagnostic skills are clearly not up to the task.
What kind of m/b is this, and what are you running on it? We've had that happen occasionally on the 48 and 64 core Penguins, which are all
The computer is a DakTEch Freedom 4 P4 DDR System. The system board is a D865GBFL w/LAN,audio & video Processor Intel Pentium 4 3.2 Ghz 800FSB I got it in 2006.
<snip> It is getting long in the tooth. (I'm about to replace my m/b of the same vintage at home.)
Interesting thought: if you go into the BIOS, there's no system event log, is there? That's normally a server thing, but who knows....
I can't imagine how you killed a video card by installing a hard drive.
Next question: is the system in a room with HVAC? Have you tried opening it up, and vacuuming dust out of the power supply and anywhere else (like the video card)?
I remember seeing a picture online a few years ago, where someone cut open a power supply, and found a lizard, too large to get out....
mark
On Mon, 25 Nov 2013, m.roth@5-cent.us wrote:
Michael Hennebry wrote:
The computer is a DakTEch Freedom 4 P4 DDR System. The system board is a D865GBFL w/LAN,audio & video Processor Intel Pentium 4 3.2 Ghz 800FSB I got it in 2006.
<snip> It is getting long in the tooth. (I'm about to replace my m/b of the same vintage at home.)
Interesting thought: if you go into the BIOS, there's no system event log, is there? That's normally a server thing, but who knows....
I've never seen anything there that looked like any kind of a log.
I can't imagine how you killed a video card by installing a hard drive.
I suppose static.
Next question: is the system in a room with HVAC? Have you tried opening it up, and vacuuming dust out of the power supply and anywhere else (like the video card)?
No and no. It's in my house in a bedroom I use for an office. Residential AC through a power strip/surge protector.
I suppose I could try dust-busting. Might be able to do it without zapping anything.
I remember seeing a picture online a few years ago, where someone cut open a power supply, and found a lizard, too large to get out....
The five year half life-someone mentioned is scary.
I'm going to be off-line for a while. Duty calls.
This time when it died, the green light on my monitor stayed green, but the screen went black. I could not switch virtual terminals or kill the server.
Examine the electrolytic capacitors on the motherboard... are any of them bulging on their tops? Or leaking oily/tar-like gunk?
Hey Michael,
I would try to check it up from bottom up and note that each time it fails you may have an error popping out later.
The first thing is to check voltage in the BIOS. Then if it's by percentage 12V should be between 11.9 to 12.1 when these are quite not the best thing to have if possible. Also take a look at the 3V and 5V to make sure that all the voltage in the machine is in the acceptable percentage which should be about 1-3% off the 12.0 3.0 5.0 ( It should be there in the BIOS)
The next step is to verify that the memory is not in high "performance" settings which can be high voltage or unverified settings. Most D865GBFL should work with most memory chips and cards out of the box. I do not remember if these boards do have memory settings in jumpers but since it's a P4 I would assume it's possible to see those (not yet finished to read the whole 142 pdf).
Try to adjust the agp Aperture size to lower then 64MB (16).
In the Hardware monitoring try to see what is the CPU heat which should be up to 70C but in some cases will show 90C+ but it's due to sensor failure.
The next step will be to restore the bios defaults settings and disabling the 1.44 (unless you have one).
This is the BIOS level I can think about from the product guide side.
The next step is to make sure you have backups for what you need (just as a regular basis task that should be done)
Also don't be tempted to replace this beast with a ARM\ATOM or any other suggestion that might not understand what a 3.2 P4 can do that the BEST ATOM cpu cannot.
I do not know where you live at and there-for the price can vary from one place to another and which can be over 200$ and over 300$.
This machine is not described as Linux compatible by INTEL and which can or cannot be a reason for anything and the change of Plug And Play flag in the bios might help to solve some problems\issues.
It is possible that the power supply was a bit loaded using two disk devices and which can cause some system freezes when a high load is there on it for a long period of time.
To make sure that the power supply is there and working properly not harming any hardware you should open the case (if it's an easy to open one) while it's off the network grid and make sure that all capacitors are in a good shape.
This is a point which you should understand this beast is old and since it works on 3.2 Ghz some parts might have gotten old but not necessarily needs to be replaced.
In a case you are replacing anything you should take couple parts together: CPU RAM Power Supply. Fans.
Sometimes it can sound a drastic change but it is recommended since there are couple unknowns in the picture which I would prefer to not discover as a fact.
I am almost sure that this CPU is a 32bit and if you don't need(like many) the fancy GRAPHICS and some additions that was added to the latest and shiny releases of Fedora then 14 is just fine.
On the next fedora release I would like to hear from someone there how many times in 5 years he replaced his chairs or his drill for example. (I assume it was not done 5 times over all these 5 years)
You can look up on the software level 4-5 times but still each time the machine got stuck some information was not written to the FS and it happens while sometime causes a problem to read a file.
The basic badblocks tool can help you discover if there is a problem with the software accessing any of the drives. Note that it happens that access to a DISK can be because of a cable sometimes.
In a case you want to make sure that the problem is in another level then the DISK you can try to work with a LIVE dvd\cd not touching any DISK IO while working on the PC.(this machine do not have USB boot support the last time I checked).
I do hope it will help you to find the right path with your PC.
Regards, Eliezer
On 25/11/13 18:58, Michael Hennebry wrote:
The computer is a DakTEch Freedom 4 P4 DDR System. The system board is a D865GBFL w/LAN,audio & video Processor Intel Pentium 4 3.2 Ghz 800FSB I got it in 2006. I switched to CentOS because Fedora will not install on it any more. Fedora 14 is the last I was able to install. Installation has almost always been a tremendous hassle for me, so I've usually not gone with the latest and greatest until my current nears EOL. I've read that a kernel bug is the reason that I could not install F16. Supposedly it had been fixed by F17, but no go.
On Thu, 28 Nov 2013, Eliezer Croitoru wrote:
The first thing is to check voltage in the BIOS. Then if it's by percentage 12V should be between 11.9 to 12.1 when these are quite not the best thing to have if possible. Also take a look at the 3V and 5V to make sure that all the voltage in the machine is in the acceptable percentage which should be about 1-3% off the 12.0 3.0 5.0 ( It should be there in the BIOS)
'Tis there. Don't remember what it said, except that 12v was 12.0v.
The next step is to verify that the memory is not in high "performance" settings which can be high voltage or unverified settings. Most D865GBFL should work with most memory chips and cards out of the box. I do not remember if these boards do have memory settings in jumpers but since it's a P4 I would assume it's possible to see those (not yet finished to read the whole 142 pdf).
The one I found is 98 pages.
Try to adjust the agp Aperture size to lower then 64MB (16).
I saw something about aperature size. Is it how many memory addresses allocated to AGP?
The next step will be to restore the bios defaults settings and disabling the 1.44 (unless you have one).
That means the floppy drive? I have one.
Also don't be tempted to replace this beast with a ARM\ATOM or any other suggestion that might not understand what a 3.2 P4 can do that the BEST ATOM cpu cannot.
Not tempted. My machine was somewhat high-end when I got it.
I do not know where you live at and there-for the price can vary from one place to another and which can be over 200$ and over 300$.
Fargo, ND.
This machine is not described as Linux compatible by INTEL and which can or cannot be a reason for anything and the change of Plug And Play flag in the bios might help to solve some problems\issues.
The current setting, which I think is the default, is Plug and Play OS no, which meands that the BIOS configures things instead of the OS.
It is possible that the power supply was a bit loaded using two disk devices and which can cause some system freezes when a high load is there on it for a long period of time.
The second drive has been there for a long time. It might have even been an option when I first got the machine. There are still two empty slots on the rack.
To make sure that the power supply is there and working properly not harming any hardware you should open the case (if it's an easy to open
I have no basis for comparing with other desktops, but I can see that I will need to open it.
one) while it's off the network grid and make sure that all capacitors are in a good shape.
I am almost sure that this CPU is a 32bit and if you don't need(like
Correct.
many) the fancy GRAPHICS and some additions that was added to the latest and shiny releases of Fedora then 14 is just fine.
Even if F14 still got security updates, I'd still want to know why CentOS has been crapping out on me. It might affect F14 eventually.
The basic badblocks tool can help you discover if there is a problem with the software accessing any of the drives. Note that it happens that access to a DISK can be because of a cable sometimes.
I got a bunch of orphan node once, but since then, fsck has been giving the partition a clean bill of health.
In a case you want to make sure that the problem is in another level then the DISK you can try to work with a LIVE dvd\cd not touching any DISK IO while working on the PC.(this machine do not have USB boot support the last time I checked).
Also did you check the Health Status in the BIOS, was the temperature in spec with what the chip spec is. You can run cpuburn (program) and see if running the processor full out causes it to shutdown.
On 11/28/2013 02:17 AM, Michael Hennebry wrote:
On Thu, 28 Nov 2013, Eliezer Croitoru wrote:
The first thing is to check voltage in the BIOS. Then if it's by percentage 12V should be between 11.9 to 12.1 when these are quite not the best thing to have if possible. Also take a look at the 3V and 5V to make sure that all the voltage in the machine is in the acceptable percentage which should be about 1-3% off the 12.0 3.0 5.0 ( It should be there in the BIOS)
'Tis there. Don't remember what it said, except that 12v was 12.0v.
The next step is to verify that the memory is not in high "performance" settings which can be high voltage or unverified settings. Most D865GBFL should work with most memory chips and cards out of the box. I do not remember if these boards do have memory settings in jumpers but since it's a P4 I would assume it's possible to see those (not yet finished to read the whole 142 pdf).
The one I found is 98 pages.
Try to adjust the agp Aperture size to lower then 64MB (16).
I saw something about aperature size. Is it how many memory addresses allocated to AGP?
The next step will be to restore the bios defaults settings and disabling the 1.44 (unless you have one).
That means the floppy drive? I have one.
Also don't be tempted to replace this beast with a ARM\ATOM or any other suggestion that might not understand what a 3.2 P4 can do that the BEST ATOM cpu cannot.
Not tempted. My machine was somewhat high-end when I got it.
I do not know where you live at and there-for the price can vary from one place to another and which can be over 200$ and over 300$.
Fargo, ND.
This machine is not described as Linux compatible by INTEL and which can or cannot be a reason for anything and the change of Plug And Play flag in the bios might help to solve some problems\issues.
The current setting, which I think is the default, is Plug and Play OS no, which meands that the BIOS configures things instead of the OS.
It is possible that the power supply was a bit loaded using two disk devices and which can cause some system freezes when a high load is there on it for a long period of time.
The second drive has been there for a long time. It might have even been an option when I first got the machine. There are still two empty slots on the rack.
To make sure that the power supply is there and working properly not harming any hardware you should open the case (if it's an easy to open
I have no basis for comparing with other desktops, but I can see that I will need to open it.
one) while it's off the network grid and make sure that all capacitors are in a good shape. I am almost sure that this CPU is a 32bit and if you don't need(like
Correct.
many) the fancy GRAPHICS and some additions that was added to the latest and shiny releases of Fedora then 14 is just fine.
Even if F14 still got security updates, I'd still want to know why CentOS has been crapping out on me. It might affect F14 eventually.
The basic badblocks tool can help you discover if there is a problem with the software accessing any of the drives. Note that it happens that access to a DISK can be because of a cable sometimes.
I got a bunch of orphan node once, but since then, fsck has been giving the partition a clean bill of health.
In a case you want to make sure that the problem is in another level then the DISK you can try to work with a LIVE dvd\cd not touching any DISK IO while working on the PC.(this machine do not have USB boot support the last time I checked).
On Mon, Nov 25, 2013 at 12:25 PM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
On the other, if it's something else, my diagnostic skills are clearly not up to the task.
Sounds like a job for DTrace http://books.google.com.ar/books?id=jseJ56fUjJgC&printsec=frontcover
DTrace Tutorial for Oracle Linux 6: http://docs.oracle.com/cd/E37670_01/E50705/html/index.html … Dynamic Tracing Guide: http://docs.oracle.com/cd/E37670_01/E38608/html/index.html …
Sorry to mention the competition on this list but OracleLinux 6.5 with UEK includes DTrace built in.
ISOs http://mirrors.wimmekes.net/pub/iso/ Public-yum http://public-yum.oracle.com/repo/OracleLinux/OL6/5/base/x86_64/
If it helps debug CentOS in your system it'd be worth it. ;)
*hides under a big rock* ;-) FC
On Thu, 5 Dec 2013, Fernando Cassia wrote:
On Mon, Nov 25, 2013 at 12:25 PM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
On the other, if it's something else, my diagnostic skills are clearly not up to the task.
Sounds like a job for DTrace
Looks interesting and useful, but I have no clue how I would apply it to the task at hand.
I've got a side of the case off. All the fans that are there are turning. There is a place for a front fan, but no fan there. Unless someone stole a fan, that is not new. The plastic front of the case would have pretty much blocked it. No obvious leakage from anything. The power supply is a sealed unit, Enlight Corporation. Cannot even look inside.
Any ideas?
In case someone finds them inspiring, I'm going to get my camera and take some pictures.
On 12/6/2013 16:34, Michael Hennebry wrote:
There is a place for a front fan, but no fan there.
It's common for cheap PC cases to have places for fans that the final PC manufacturer chooses not to populate.
If you decide to put a fan there, be sure to orient it so it blows "in line" with the case exhaust fans. You don't want two fans blowing opposite directions on opposite sides of the case. The plastic fan shroud usually has a molded-in arrow showing the airflow direction.
The plastic front of the case would have pretty much blocked it.
Maybe not. Some PC case bezels have vents at the bottom edge. A common pattern is for the front PC fan to suck air in, and for the rear PC fans to blow air out. Since they're usually at opposite corners of the case, this forces air to flow through the entire case.
The power supply is a sealed unit,
I think you'll find that once you unscrew it from the case, you'll expose another set of screws that will let you remove the power supply's lid. The odd hole in the back of the case is designed to block access to these screws, on purpose.
Don't touch anything in there unless you know your microfarads from your microhenries. Just take pics.
I'm going to get my camera and take some pictures.
Please do. We may well see something you didn't.
Some advice, based on prior experience receiving uselessly bad pictures in the DIY electronics slice of my life:
1a. Turn on lots of lights and shine them into the case. Experiment with forced camera flash. Electronics enclosures (including PC cases) are often dark places, which means not enough photons for your camera to take a fast, sharp picture. If you can't get enough artificial light into the case, take it outside and shoot into the case with the sun over your shoulder.
1b. Bounce or diffuse as much of the light as possible. Lots of direct light is good, but if it creates blown-out flare spots or inky shadows that obscure detail, it's still no good. There are many ways to make cheap diffusers and bounce cards: old thin sheets, tin foil, poster board... Tenting a sheet over your head and the case can give a better result than a bright direct light. If your camera's flash is articulated, bounce it into the scene rather than shoot directly in.
2. Use your camera's macro function, if it has one. 10 separate pictures of 10 details is better than one overall picture where you can't even tell how many pins are on a given chip.
3. Use a tripod, if you have one. If you don't, brace the camera against something: a nearby wall, the PC case, a sandbag... A camera on a tripod set for a 30 second exposure can compensate for a *lot* of problems in area #1.
On 12/6/2013 4:39 PM, Warren Young wrote:
On 12/6/2013 16:34, Michael Hennebry wrote:
There is a place for a front fan, but no fan there.
It's common for cheap PC cases to have places for fans that the final PC manufacturer chooses not to populate.
its pretty common on high end cases too. the one I have now has room for like 8 120mm fans, of which i'm using 3.
On 12/6/2013 18:00, John R Pierce wrote:
On 12/6/2013 4:39 PM, Warren Young wrote:
On 12/6/2013 16:34, Michael Hennebry wrote:
There is a place for a front fan, but no fan there.
It's common for cheap PC cases to have places for fans that the final PC manufacturer chooses not to populate.
its pretty common on high end cases too. the one I have now has room for like 8 120mm fans, of which i'm using 3.
The only reason I phrased it that way is that a lot of high-end machines use bespoke cases, where every fan position is populated because it was designed together as a system.
Anything built from a collection of off-the-shelf parts, as DIYers and some of the smaller high-end PC makers do, can have this same sort of design mismatch.
On Fri, 6 Dec 2013, Warren Young wrote:
On 12/6/2013 16:34, Michael Hennebry wrote:
The power supply is a sealed unit,
I think you'll find that once you unscrew it from the case, you'll expose another set of screws that will let you remove the power supply's lid. The odd hole in the back of the case is designed to block access to these screws, on purpose.
If I remove the right screws, I think that moving it farther into the case would allow it to be removed. For that, I'd want it on its side so that it didn't fall on something.
Don't touch anything in there unless you know your microfarads from your microhenries. Just take pics.
I do, but I'm still not likely to try to fix it.
Some advice, based on prior experience receiving uselessly bad pictures in the DIY electronics slice of my life:
1b. Bounce or diffuse as much of the light as possible. Lots of direct light is good, but if it creates blown-out flare spots or inky shadows that obscure detail, it's still no good. There are many ways to make cheap diffusers and bounce cards: old thin sheets, tin foil, poster board... Tenting a sheet over your head and the case can give a better result than a bright direct light. If your camera's flash is articulated, bounce it into the scene rather than shoot directly in.
My office has a large flourescent light. If the pictures are not well enough lit, I can change the illumination angle.
I'm open to suggestion, both in regard to lighting and in regard to subject.
For now, nap time.
On 12/6/2013 7:02 PM, Michael Hennebry wrote:
If I remove the right screws, I think that moving it farther into the case would allow it to be removed. For that, I'd want it on its side so that it didn't fall on something.
nearly all PC power supplies are standard ATX/EPS format. remove side panel, case on side, unplug ALL the power wires, untangle and extract the power harness, remove the 4 screws on the back panel around the PSU vents, and it lifts out from the inside of the case. replace with a decent brand supply of equal or higher wattage rating, and the correct ATX version (v2.1 or whatever). really old systems used much more 5V than modern PSU's offer, and used less 12V. modern PSU's are mostly all about the 12V, the other voltages are lower current. many newer systems need additional 12V plugs for the mainboard, like 2x2 or 2x3 connectors.
In some system cases, the PSU is screwed to a shelf or bracket or back plate that comes out with it as a unit, then you disassemble them.
I'd thought I'd sent this already. Apparently the last crap out was before I'd hit send. BTW the last few crap outs have been followed with processor area temps up to 60 C.
I suspect that my jpg's are too big to go through as attachments, so I've put them on my web site: http://web.cs.ndsu.nodak.edu/~hennebry/computer/
I do not recognize much, but I will list things I do. The board with the blue heat sink is my video card. The box in the upper left is the power supply. The four card between the big black Intel fan and the top storage rack is my RAM. The top storage rack containg a CD reader on top and a DVR-R burner just below it. The lower storage rack contains a floppy drive on top and two hard drives below it. I do not remember which is the IDE and which is the new SATA. Below the empty fan case below the storage racks is a pair of USB ports. I suspect the pincushiony thing between the video card and the big black Intel fan of being the heat sink for the CPU, but I do not know.
On 07 December 2013 @02:57 zulu, Michael Hennebry wrote:
I do not remember which is the IDE and which is the new SATA.
The SATA drive has the thin red cable connecting it to the motherboard.
I suspect the pincushiony thing between the video card and the big black Intel fan of being the heat sink for the CPU, but I do not know.
That's the heat sink for the northbridge chipset.
The CPU heat sink is under the fan pointing down towards the motherboard. You lift those 2 levers to release it, and there's likely another lever under it all locking the CPU into the socket.
The only electrolytics I see on the motherboard that might be swollen are in the row right next to the CPU, towards the fan blowing out the back. But it's hard to see the tops of them, too... the 'K' cut into the tops of the electrolytics are where they're intended to rupture (instead of exploding like blasting caps)... not unlike the way cement workers score the surface after troweling, to control where sidewalks/driveways eventually crack.
The thin grey cables connecting the CD and DVD drives to the motherboard are most-likely not needed... the audio signal should go through the IDE ribbon cables without those.
On Sat, 2013-12-07 at 00:58 -0500, Darr247 wrote:
On 07 December 2013 @02:57 zulu, Michael Hennebry wrote:
The CPU heat sink is under the fan pointing down towards the motherboard. You lift those 2 levers to release it, and there's likely another lever under it all locking the CPU into the socket.
The picture is not very clear, but it looks as if the heatsink on the processor has collected a lot of dust, probably blocking the air flow. Do not yet remove the heat sink ( I am always weary of doing so) but try to remove the dust first if there is a lot of it. Louis
On 12/7/2013 3:01 AM, Louis Lagendijk wrote:
On Sat, 2013-12-07 at 00:58 -0500, Darr247 wrote:
On 07 December 2013 @02:57 zulu, Michael Hennebry wrote:
The CPU heat sink is under the fan pointing down towards the motherboard. You lift those 2 levers to release it, and there's likely another lever under it all locking the CPU into the socket.
The picture is not very clear, but it looks as if the heatsink on the processor has collected a lot of dust, probably blocking the air flow. Do not yet remove the heat sink ( I am always weary of doing so) but try to remove the dust first if there is a lot of it.
geez, its a 10 year old pentium-4, and not even a late p4, a middle aged 32bit-only one.... put it out of its misery, its been living on borrowed time for the last 5 years.
On Sat, 7 Dec 2013, John R Pierce wrote:
geez, its a 10 year old pentium-4, and not even a late p4, a middle aged 32bit-only one.... put it out of its misery, its been living on borrowed time for the last 5 years.
You mean since it was two?
Even when I have income, I get annoyed about little mysteries that make things just not work. I do not have any income, so I also get annoyed at these little expenses that pop up occasionaly. Apparently just to annoy me.
The symptoms have been changing. Instead of a black screen, after a crap out, I usually get a corduroy (sp?) pattern composed of different darknesses of one color.
Sometimes pushing the reset button brings it back. Often I have to do a power-off. When I turn the power back on, the machine does not boot, the reset button still needs pressing.
All this before I decided to do some dusting. Turning it one one last time before dusting, I noticed that the video card fan was not spinning. Whether that was a change or I'd been inobservant previously, I do not know.
In light of previous comments about just moving dust around, the Dust-Off started at the top and moved down. It blew chunks out of the video card fan.
The video card fan spins now, but symptoms have not changed since the dusting. I can only use CentOS for several minutes. F14 will usually last a lot longer.
On Sun, Dec 8, 2013 at 10:27 AM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
Even when I have income, I get annoyed about little mysteries that make things just not work. I do not have any income, so I also get annoyed at these little expenses that pop up occasionaly. Apparently just to annoy me.
You are taking this way too personally. Things don't get old and break just to annoy you - it happens to everyone. Realistically, even if you fix the power supply or whatever the cause is this time, it won't be long before something else goes. That refurb box would at least advance the odds a few years.
On Sun, 8 Dec 2013, Michael Hennebry wrote:
On Sat, 7 Dec 2013, John R Pierce wrote:
geez, its a 10 year old pentium-4, and not even a late p4, a middle aged 32bit-only one.... put it out of its misery, its been living on borrowed time for the last 5 years.
You mean since it was two?
Even when I have income, I get annoyed about little mysteries that make things just not work. I do not have any income, so I also get annoyed at these little expenses that pop up occasionaly. Apparently just to annoy me.
Perhaps John meant Get a Real Computer.
On Sun, 8 Dec 2013, Les Mikesell wrote:
On Sun, Dec 8, 2013 at 10:27 AM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
Even when I have income, I get annoyed about little mysteries that make things just not work. I do not have any income, so I also get annoyed at these little expenses that pop up occasionaly. Apparently just to annoy me.
You are taking this way too personally. Things don't get old and break just to annoy you - it happens to everyone. Realistically, even if you fix the power supply or whatever the cause is this time, it won't be long before something else goes. That refurb box would at least advance the odds a few years.
Realistically, my abilities and my income circumscribe what I can and what I should do.
On 12/8/2013 8:27 AM, Michael Hennebry wrote:
On Sat, 7 Dec 2013, John R Pierce wrote:
geez, its a 10 year old pentium-4, and not even a late p4, a middle aged 32bit-only one.... put it out of its misery, its been living on borrowed time for the last 5 years.
You mean since it was two?
i find most computer electronics have a half life of about 5 years. after 5 years, they get increasingly flakey. your motherboard is a 10 year old system, hence my 'borrowed time for 5 years' statement.
the 865G chipset on that motherboard was new in 2003, and the Pentium-4 "northwood" CPU (I think thats what you have) were obsoleted by 2004 (processors started coming in socket 775 rather than 478 circa summer 2004).
that D865GBF motherboard was new in April 2003, and the final specification update was November 2004, although they probably did sell it for another year or so before it was withdrawn. It was taken off support in 2007, so if you got it 7 years ago (2006?), it was already nearing its end-of-support-life.
On Sun, 8 Dec 2013, John R Pierce wrote:
On 12/8/2013 8:27 AM, Michael Hennebry wrote:
On Sat, 7 Dec 2013, John R Pierce wrote:
geez, its a 10 year old pentium-4, and not even a late p4, a middle aged 32bit-only one.... put it out of its misery, its been living on borrowed time for the last 5 years.
You mean since it was two?
i find most computer electronics have a half life of about 5 years. after 5 years, they get increasingly flakey. your motherboard is a 10 year old system, hence my 'borrowed time for 5 years' statement.
You mean it was rotting just sitting on a shelf? That is assuming mine was one of the first ones made.
the 865G chipset on that motherboard was new in 2003, and the Pentium-4 "northwood" CPU (I think thats what you have) were obsoleted by 2004 (processors started coming in socket 775 rather than 478 circa summer 2004).
that D865GBF motherboard was new in April 2003, and the final specification update was November 2004, although they probably did sell it for another year or so before it was withdrawn. It was taken off support in 2007, so if you got it 7 years ago (2006?), it was already nearing its end-of-support-life.
"The warranty wore out."
On 12/08/2013 03:31 PM, Michael Hennebry wrote:
You mean it was rotting just sitting on a shelf?
Perhaps surprisingly, systems of that age *can* fairly literally rot. There were a number of Taiwanese electrolytic capacitor manufacturers that "borrowed" a partial recipe from a Japanese company: One that was unfortunately missing an important component that kept the paste from eating the capacitor from inside out. It often initially manifested as system instability.
See: https://en.wikipedia.org/wiki/Capacitor_plague
I lost a couple of motherboards to it.
I've tried to open the power supply to look for capacitor guts and such. I've unscrewed it from the case and removed the screws that seemed sufficient for separation. The stumbling block seems to be under the "QC PASS" sticker near the rainbow of wires coming out. I've gotten the edge of a box cutter in the gap that had been covered by the sticker, so I know that the sticker itself is not the problem.
See web.cs.ndsu.nodak.edu/~hennebry/computer/ps[123].jpg .
Any ideas on how to proceed?
I noticed the the grub stanza for my F14 contains the line kernel /vmlinuz-2.6.35.14-106.fc14.i686.PAE ... For CentOS, I have kernel /boot/vmlinuz-2.6.32-358.23.2.el6.i686 ... I note the absence of PAE for CentOS. I've read that PAE can be important. Could it be the reason that an F14 boot lasts so much longer than CentOS? If so, how do I get a PAE kernal for CentOS?
If not, could I get or make a kernel version for CentOS similar to the one F14 uses?
On 12/13/2013 16:35, Michael Hennebry wrote:
I note the absence of PAE for CentOS. I've read that PAE can be important.
Only if you're trying to address more than 4 GB of RAM on a 32-bit system. Even then, most software doesn't take advantage of it.
PAE is an old hack Intel invented in the mid 90's to ease the transition to 64-bit computing. Now that 64-bit is really, finally here, it's time to start forgetting that PAE ever existed.
On Mon, Dec 16, 2013 at 11:28 AM, Warren Young warren@etr-usa.com wrote:
On 12/13/2013 16:35, Michael Hennebry wrote:
I note the absence of PAE for CentOS. I've read that PAE can be important.
Only if you're trying to address more than 4 GB of RAM on a 32-bit system. Even then, most software doesn't take advantage of it.
PAE is an old hack Intel invented in the mid 90's to ease the transition to 64-bit computing. Now that 64-bit is really, finally here, it's time to start forgetting that PAE ever existed.
I think PAE is a required default in CentOS6 instead of an option. CentOS5 had separate kernel versions.
On 12/14/2013 12:38 PM, Warren Young wrote:
On 12/13/2013 16:20, Michael Hennebry wrote:
Any ideas on how to proceed?
There may be a final screw under one of the stickers.
Yes, it looks to me like there should be a screw under the QC Pass sticker, right under the round circle.
Peter
On Sat, 14 Dec 2013, Peter wrote:
On 12/14/2013 12:38 PM, Warren Young wrote:
On 12/13/2013 16:20, Michael Hennebry wrote:
Any ideas on how to proceed?
There may be a final screw under one of the stickers.
Yes, it looks to me like there should be a screw under the QC Pass sticker, right under the round circle.
Of course. Thanks. I'll check it out on Sunday.
On Sat, 14 Dec 2013, Peter wrote:
On 12/14/2013 12:38 PM, Warren Young wrote:
On 12/13/2013 16:20, Michael Hennebry wrote:
Any ideas on how to proceed?
There may be a final screw under one of the stickers.
Yes, it looks to me like there should be a screw under the QC Pass sticker, right under the round circle.
Bullseye. I took more pictures and added web.cs.ndsu.nodak.edu/~hennebry/computer/ps[45].jpg to my website.
No bulging capacitors, but the yellow stuff looks suspicious. All the yellow stuff is visible in the pictures. The darkness near the smaller yellow spot indicates my inability to position both the camera and the light the way I wanted them.
Can anyone tell me for sure whether that yellow stuff is supposed to not be there?
On 12/16/2013 12:21 PM, Michael Hennebry wrote:
On Sat, 14 Dec 2013, Peter wrote:
On 12/14/2013 12:38 PM, Warren Young wrote:
On 12/13/2013 16:20, Michael Hennebry wrote:
Any ideas on how to proceed?
There may be a final screw under one of the stickers.
Yes, it looks to me like there should be a screw under the QC Pass sticker, right under the round circle.
Bullseye. I took more pictures and added web.cs.ndsu.nodak.edu/~hennebry/computer/ps[45].jpg to my website.
#4 gives permission denied.
No bulging capacitors, but the yellow stuff looks suspicious. All the yellow stuff is visible in the pictures. The darkness near the smaller yellow spot indicates my inability to position both the camera and the light the way I wanted them.
Can anyone tell me for sure whether that yellow stuff is supposed to not be there?
I can't say for sure, but I think it's just glue that's used during the assembly process.
Peter
On Mon, 16 Dec 2013, Peter wrote:
On 12/16/2013 12:21 PM, Michael Hennebry wrote:
I took more pictures and added web.cs.ndsu.nodak.edu/~hennebry/computer/ps[45].jpg to my website.
#4 gives permission denied.
Fixed now. Don't know why it was the only one without read permission. Copied ps4.jpg and ps5.jpg with the same command.
No bulging capacitors, but the yellow stuff looks suspicious. All the yellow stuff is visible in the pictures. The darkness near the smaller yellow spot indicates my inability to position both the camera and the light the way I wanted them.
Can anyone tell me for sure whether that yellow stuff is supposed to not be there?
I can't say for sure, but I think it's just glue that's used during the assembly process.
Looks like I'll need to see whether I still have my multimeter.
On 12/16/2013 12:49 PM, Michael Hennebry wrote:
Looks like I'll need to see whether I still have my multimeter.
Not sure what you're planning to find on your multimeter, bad caps are hard to detect. The only real way I know of to properly test if a problem is originating from a PSU is to swap it with a known good one.
At any rate, I would be very careful with the mm, a charged cap can damage it.
Peter
I see 2 components in ps4.jpg that look like they've ruptured.
One in the mid/foreground with the yellow hot glue on it (the shorter one, between the inductor and the caps), and one hiding under the harness that exits the supply... to the upper-right of the green cap, near the PS housing.
On Sun, 15 Dec 2013, Darr247 wrote:
I see 2 components in ps4.jpg that look like they've ruptured.
One in the mid/foreground with the yellow hot glue on it (the shorter
The one with the visible VENT and 105 printing?
one, between the inductor and the caps), and one hiding under the harness that exits the supply... to the upper-right of the green cap, near the PS housing.
And to the right of the pink padlocky thing. To me it looks too fuzzy to tell. I'm sure that is partly the dust.
VENT105 does look like it's bulging a bit. I'll take another look to make sure it isn't a camera thing.
On 12/16/2013 09:53, Michael Hennebry wrote:
The one with the visible VENT and 105 printing?
"Vent" just calls out that there is a vent on the top of the cap, which it obvious without the label. It's the scoring in the metal, which allows the top of the cap to break open in a controlled way if the pressure inside gets too high. Without the vent, a failing cap holds the pressure in until it explodes like a firecracker.
The other label is actually 105℃, meaning that it's rated for a certain number of hours of use at 105℃. That means it's probably a fairly high quality cap. If you have to replace this cap, you'd want to match that temperature rating as well as the capacitance and voltage rating.
You'll probably find that *all* of the caps are 105℃ rated, so they'll have that marking on them somewhere.
(You think things are complicated already? Wait until you start shopping for caps!)
And to the right of the pink padlocky thing.
??
Do you mean the chokes, which have yellow cores and red and orange lacquered wire?
Or perhaps the dark gray cored inductors, with red wire?
Or do you mean the pink resistors(?) near the board edge?
To me it looks too fuzzy to tell. I'm sure that is partly the dust.
Yes. You should have blown the dust out before taking the pics. It needs to be blown out regardless.
VENT105 does look like it's bulging a bit.
It doesn't look like it to me.
On 16 December 2013 @16:53 zulu, Michael Hennebry wrote:
On Sun, 15 Dec 2013, Darr247 wrote:
I see 2 components in ps4.jpg that look like they've ruptured.
One in the mid/foreground with the yellow hot glue on it (the shorter
The one with the visible VENT and 105 printing?
No. The short one (in 'front' of the electrolytic capacitor with the markings you mention) that appears to have an output choke (inductor), wound around the component and likely terminated to each lead.
The other one is to the right of the pink resistor, yes... it *also* appears to have an output choke wound around it.
Those are quite unusual, in my experience. I just took apart 3 dead power supplies here in the recycle bin and found swollen caps in all of them, but no chokes wound around components like that. Typically, inductor coils are wound around a piece of ferrous metal to intensify the flux, amplifying their filtering effect. On one hand you'd think it's a design flaw that extra chokes would need to be added like that to specific components; on the other hand it would point to extra engineering having been done to actually fix a specific problem instead of just pushing them out the door.
Then again - as others have mentioned - maybe it's just clumps of dust... and I'm imagining things. :)
On 12/15/2013 16:49, Michael Hennebry wrote:
Looks like I'll need to see whether I still have my multimeter.
That's not likely to tell you much.
About the only thing I'd trust a typical DMM to tell me about a PSU is whether its rails are within voltage spec. You must do that test under load, and you need a fairly accurate DMM to get a result you can trust.
If the PSU passes the voltage test, it could still be bad. The only way to tell, short of just swapping it, is to do an ESR test[*], which requires an LCR meter. (It also requires removing the PSU board from the box, which exposes you to the dangers of AC wiring and charged caps.)
I don't think I've ever seen a DMM with an integrated ESR function. If such a thing does exist, the DMM would end up being pretty expensive if it gave results you could trust for this, since we're talking about measurements under 1 ohm. Decent LCR meters start at about $200 and go way up from there. Cheaper to just gamble on a PSU swap.
[*] https://en.wikipedia.org/wiki/Equivalent_series_resistance
On 12/15/2013 16:21, Michael Hennebry wrote:
the yellow stuff looks suspicious.
It's a kind of strain relief. Without that flexible glue, dropping the computer could snap those caps off at their base. Since this is the sort of thing that occasionally happens to computers in shipping, computer manufacturers try to make sure their machines can withstand a few of these sharp shocks.
After all, damage in shipping is damage during the warranty period, and the shipping companies are a PITA to get insurance money out of.
The darkness near the smaller yellow spot indicates my inability to position both the camera and the light the way I wanted them.
That's why I mentioned reflectors and bounce cards among my photo advice.
On Fri, Dec 13, 2013 at 6:20 PM, Michael Hennebry < hennebry@web.cs.ndsu.nodak.edu> wrote:
I've tried to open the power supply to look for capacitor guts and such.
If there was a problem with the PSU, you would experience it in your F14 install you spoke of.
I've unscrewed it from the case and removed the screws that seemed sufficient for separation. The stumbling block seems to be under the "QC PASS" sticker near the rainbow of wires coming out. I've gotten the edge of a box cutter in the gap that had been covered by the sticker, so I know that the sticker itself is not the problem.
See web.cs.ndsu.nodak.edu/~hennebry/computer/ps[123].jpg .
Any ideas on how to proceed?
Rather than rip the PSU open (and hope you don't get zapped good by a charged capacitor...) just hook a power supply tester to it. Or look online for instructions on testing it with a multimeter (more tedious than the PSU tester).
-- Michael hennebry@web.cs.ndsu.NoDak.edu "Friday the Thirteenth was always unlucky." -- Mr. Peabody _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
On Fri, 13 Dec 2013, SilverTip257 wrote:
On Fri, Dec 13, 2013 at 6:20 PM, Michael Hennebry < hennebry@web.cs.ndsu.nodak.edu> wrote:
I've tried to open the power supply to look for capacitor guts and such.
If there was a problem with the PSU, you would experience it in your F14 install you spoke of.
F14 craps out also. F14 just takes a lot longer.
Rather than rip the PSU open (and hope you don't get zapped good by a charged capacitor...) just hook a power supply tester to it. Or look online for instructions on testing it with a multimeter (more tedious than the PSU tester).
I'll look, not touch.
On 12/13/2013 7:21 PM, Michael Hennebry wrote:
Rather than rip the PSU open (and hope you don't get zapped good by a charged capacitor...) just hook a power supply tester to it. Or look online for instructions on testing it with a multimeter (more tedious than the PSU tester).
I'll look, not touch.
me, if I had any reason to suspect a power supply, I would just get a new one, basic PC power supplies in reasonable wattage ratings are quite cheap. examples: http://www.newegg.com/Product/Product.aspx?Item=N82E16817152032 http://www.newegg.com/Product/Product.aspx?Item=N82E16817371003
On Sat, 14 Dec 2013, John R Pierce wrote:
me, if I had any reason to suspect a power supply, I would just get a new one, basic PC power supplies in reasonable wattage ratings are quite cheap. examples: http://www.newegg.com/Product/Product.aspx?Item=N82E16817152032 http://www.newegg.com/Product/Product.aspx?Item=N82E16817371003
From this, I infer that 20-pin ATX's are sufficiently standardized
that I do not need to be model- or brand-specific.
On 12/16/2013 8:56 AM, Michael Hennebry wrote:
From this, I infer that 20-pin ATX's are sufficiently standardized
that I do not need to be model- or brand-specific.
well, ATX 1.x stuff had more 5V and less 12V, while ATX 2.x boosts the 12V output capacity and has less 5V... I'm pretty sure even a AGP/PCI P4 is ATX 2.0, but I suppose I could be wrong. We're stirring some mighty old neurons here. I ejunked all my P4's some time ago.
On 12/16/2013 10:03 AM, John R Pierce wrote:
On 12/16/2013 8:56 AM, Michael Hennebry wrote:
From this, I infer that 20-pin ATX's are sufficiently standardized
that I do not need to be model- or brand-specific.
well, ATX 1.x stuff had more 5V and less 12V, while ATX 2.x boosts the 12V output capacity and has less 5V... I'm pretty sure even a AGP/PCI P4 is ATX 2.0, but I suppose I could be wrong. We're stirring some mighty old neurons here. I ejunked all my P4's some time ago.
ok, I googled it, the manual for your motherboard says "ATX12V", which is close enough the 2.x spec that it should work. I do see a suggestion that you use a somewhat higher 'wattage' ATX12V 2.x PSU to compensate for the differences.
btw, here's the technical product manual for your mainboard http://downloadmirror.intel.com/15207/eng/D865GBF_D865GLC_ProductGuide02_Eng...
At 10:27 AM 12/16/2013, you wrote:
On 12/16/2013 10:03 AM, John R Pierce wrote:
On 12/16/2013 8:56 AM, Michael Hennebry wrote:
From this, I infer that 20-pin ATX's are sufficiently standardized
that I do not need to be model- or brand-specific.
well, ATX 1.x stuff had more 5V and less 12V, while ATX 2.x boosts the 12V output capacity and has less 5V... I'm pretty sure even a AGP/PCI P4 is ATX 2.0, but I suppose I could be wrong. We're stirring some mighty old neurons here. I ejunked all my P4's some time ago.
ok, I googled it, the manual for your motherboard says "ATX12V", which is close enough the 2.x spec that it should work. I do see a suggestion that you use a somewhat higher 'wattage' ATX12V 2.x PSU to compensate for the differences.
btw, here's the technical product manual for your mainboard http://downloadmirror.intel.com/15207/eng/D865GBF_D865GLC_ProductGuide02_Eng...
-- john r pierce 37N 122W somewhere on the middle of the left coast
Is there some reason that this conversation is in the public channel? I'm not sure that power connectors to ATX boards is really relevant to the forum's topic.
David Kurn
On Mon, 16 Dec 2013, John R Pierce wrote:
On 12/16/2013 10:03 AM, John R Pierce wrote:
On 12/16/2013 8:56 AM, Michael Hennebry wrote:
From this, I infer that 20-pin ATX's are sufficiently standardized
that I do not need to be model- or brand-specific.
well, ATX 1.x stuff had more 5V and less 12V, while ATX 2.x boosts the 12V output capacity and has less 5V... I'm pretty sure even a AGP/PCI P4 is ATX 2.0, but I suppose I could be wrong. We're stirring some mighty old neurons here. I ejunked all my P4's some time ago.
ok, I googled it, the manual for your motherboard says "ATX12V", which is close enough the 2.x spec that it should work. I do see a suggestion that you use a somewhat higher 'wattage' ATX12V 2.x PSU to compensate for the differences.
btw, here's the technical product manual for your mainboard http://downloadmirror.intel.com/15207/eng/D865GBF_D865GLC_ProductGuide02_Eng...
That is where I got the 20-pin ATX stuff. It just occured to me that I might want geometric as well as electrical compatibility.
On 12/16/2013 08:28 PM, Michael Hennebry wrote:
btw, here's the technical product manual for your mainboard http://downloadmirror.intel.com/15207/eng/D865GBF_D865GLC_ProductGuide02_Eng...
That is where I got the 20-pin ATX stuff. It just occured to me that I might want geometric as well as electrical compatibility.
I wonder if its time to start considering a social list again; very little of the conversation in this thread is really CentOS specific now.
On Mon, 16 Dec 2013, Karanbir Singh wrote:
I wonder if its time to start considering a social list again; very little of the conversation in this thread is really CentOS specific now.
On Fri, 13 Dec 2013, Michael Hennebry wrote:
I noticed the the grub stanza for my F14 contains the line kernel /vmlinuz-2.6.35.14-106.fc14.i686.PAE ... For CentOS, I have kernel /boot/vmlinuz-2.6.32-358.23.2.el6.i686 ... I note the absence of PAE for CentOS. I've read that PAE can be important. Could it be the reason that an F14 boot lasts so much longer than CentOS? If so, how do I get a PAE kernal for CentOS?
If not, could I get or make a kernel version for CentOS similar to the one F14 uses?
On Mon, 16 Dec 2013, Les Mikesell wrote:
I think PAE is a required default in CentOS6 instead of an option. CentOS5 had separate kernel versions.
Even if PAE is not important, the different kernel versions might be the reason that F14 lasts longer than CentOS 6. Could someone point me to directions for getting or making a 2.6.35... kernal for CentoS 6?
On Tue, Dec 17, 2013 at 4:55 PM, Michael Hennebry < hennebry@web.cs.ndsu.nodak.edu> wrote:
On Mon, 16 Dec 2013, Karanbir Singh wrote:
I wonder if its time to start considering a social list again; very little of the conversation in this thread is really CentOS specific now.
+1 Maybe ... :)
This thread will never die! :-S
On Fri, 13 Dec 2013, Michael Hennebry wrote:
I noticed the the grub stanza for my F14 contains the line kernel /vmlinuz-2.6.35.14-106.fc14.i686.PAE ... For CentOS, I have kernel /boot/vmlinuz-2.6.32-358.23.2.el6.i686 ... I note the absence of PAE for CentOS. I've read that PAE can be important. Could it be the reason that an F14 boot lasts so much longer than CentOS? If so, how do I get a PAE kernal for CentOS?
If not, could I get or make a kernel version for CentOS similar to the one F14
uses?
On Mon, 16 Dec 2013, Les Mikesell wrote:
I think PAE is a required default in CentOS6 instead of an option. CentOS5 had separate kernel versions.
Even if PAE is not important, the different kernel versions might be the reason that F14 lasts longer than CentOS 6. Could someone point me to directions for getting or making a 2.6.35... kernal for CentoS 6?
Easiest is probably to grab the SRPM for that kernel version and recompile it. But at the same time, you might be able to just snag the RPM for that kernel version and transplant it. CentOS6 is similar to Fedora 12 in package versions [0] so it may or may not work so well with F14 pieces.
** You are responsible for whatever dependency problems this transplanting might create. **
Here's what you asked for ... the SRPM of that F14 kernel at [2] and the non-PAE RPM at [3] and the PAE one [4] Fedora custom kernel building instructions at [1]
[0] http://fedoraproject.org/wiki/Red_Hat_Enterprise_Linux#History [1] http://fedoraproject.org/wiki/Building_a_custom_kernel [2] http://archive.fedoraproject.org/pub/archive/fedora/linux/updates/14/SRPMS/k... [3] http://archive.fedoraproject.org/pub/archive/fedora/linux/updates/14/i386/ke... [4] http://archive.fedoraproject.org/pub/archive/fedora/linux/updates/14/i386/ke...
-- Michael hennebry@web.cs.ndsu.NoDak.edu "SCSI is NOT magic. There are *fundamental technical reasons* why it is necessary to sacrifice a young goat to your SCSI chain now and then." -- John Woods _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Preparatory to removing the PSU, I've been disconnecting it. Those Molex things are tough. So far, the hardest one I've removed is the 20-pin connector from the MB. It took me days. The 4-pin connector on the MB is proving to be harder. The receptacle seems to be more firmly attached to the plug than to th MB. I've tried getting a small screwdriver between the pieces and twisting, but the plastic gives before anything else. Just getting the screwdriver into such cramped quarters was difficult.
I'm considering cutting the wires and splicing on another receptacle. I'd like a better idea. Any suggestions?
Michael Hennebry wrote:
Preparatory to removing the PSU, I've been disconnecting it. Those Molex things are tough. So far, the hardest one I've removed is the 20-pin connector from the MB. It took me days. The 4-pin connector on the MB is proving to be harder. The receptacle seems to be more firmly attached to the plug than to th MB. I've tried getting a small screwdriver between the pieces and twisting, but the plastic gives before anything else. Just getting the screwdriver into such cramped quarters was difficult.
I'm considering cutting the wires and splicing on another receptacle. I'd like a better idea. Any suggestions?
Try reading glasses or a magnifying glass (I needed the former). Then find the plastic tab that locks the things in place, and push the other end to release.
mark
On 1/15/2014 1:43 PM, Michael Hennebry wrote:
Preparatory to removing the PSU, I've been disconnecting it. Those Molex things are tough. So far, the hardest one I've removed is the 20-pin connector from the MB. It took me days. The 4-pin connector on the MB is proving to be harder. The receptacle seems to be more firmly attached to the plug than to th MB. I've tried getting a small screwdriver between the pieces and twisting, but the plastic gives before anything else.
there's a tab you depress, then those connectors should come out with just a few pounds of force.
On Wed, 15 Jan 2014, John R Pierce wrote:
On 1/15/2014 1:43 PM, Michael Hennebry wrote:
Preparatory to removing the PSU, I've been disconnecting it. Those Molex things are tough. So far, the hardest one I've removed is the 20-pin connector from the MB. It took me days. The 4-pin connector on the MB is proving to be harder. The receptacle seems to be more firmly attached to the plug than to th MB. I've tried getting a small screwdriver between the pieces and twisting, but the plastic gives before anything else.
there's a tab you depress, then those connectors should come out with just a few pounds of force.
Thank you. That did the trick. I'd seen the tab, just hadn't realized it was something useful.
'Twasn't the PSU. I replaced it and got the same symptoms.
Pardon me. I need to go kill something.
On Tue, Jan 21, 2014 at 8:00 PM, Michael Hennebry hennebry@web.cs.ndsu.nodak.edu wrote:
'Twasn't the PSU. I replaced it and got the same symptoms.
I still have a beige box here...
Pardon me. I need to go kill something.
-- Michael hennebry@web.cs.ndsu.NoDak.edu "SCSI is NOT magic. There are *fundamental technical reasons* why it is necessary to sacrifice a young goat to your SCSI chain now and then." -- John Woods _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
On 22/01/14 03:00, Michael Hennebry wrote:
'Twasn't the PSU. I replaced it and got the same symptoms.
Pardon me. I need to go kill something.
Hey Michael,
Don't run to kill something.. it will not help but it will...
There are issues related to hardware which not everyone has the tools to identify. It is one of the fundamentals that you cannot always able to do what others can.. This is a fact of life which we cannot resist. We are obligated to first make sure one thing or another in our level was checked and verified but still nothing happens.
What PSU is it the new one?
Eliezer
On Wed, 22 Jan 2014, Eliezer Croitoru wrote:
What PSU is it the new one?
The new one is still installed. Installing it was mechanically difficult. I won't put the other one back without cause. Cause would be expecting it to work. The new one is a PSU RAIDMAX | RX-380K 380W RT. All all voltages it will produce at least as much current as its predecessor. That said, it has fewer connectors. I had to leave off my CD and floppy drives.
I thought it had already been determined this was not CentOS related?
http://www.diy-computer-repair.com/ http://www.thepcmanwebsite.com/computer_repair.shtml et al
Or, answer a couple questions per month in your area of expertise and earn free membership to ask questions in other topic areas: e.g. http://www.experts-exchange.com/Hardware/Desktops/
Well a RX3xx is a very good one. Hope you will have luck with it! If you have questions feel free to post them! CD and floppy are old and indeed needed in many cases but I have machines which doesn't have these at all in to the favor of USB :\
Eliezer
On 22/01/14 20:15, Michael Hennebry wrote:
The new one is still installed. Installing it was mechanically difficult. I won't put the other one back without cause. Cause would be expecting it to work. The new one is a PSU RAIDMAX | RX-380K 380W RT. All all voltages it will produce at least as much current as its predecessor. That said, it has fewer connectors. I had to leave off my CD and floppy drives.
On 01/23/2014 10:59 PM, Peter wrote:
On 01/23/2014 07:15 AM, Michael Hennebry wrote:
That said, it has fewer connectors. I had to leave off my CD and floppy drives.
I have yet to see a power supply that doesn't have connectors for these, but you can get adapters and/or splitters for that if need be.
...in fact: http://raidmax.com/psu/rx_380k.html
it has four molex four pin connectors any one of which should be suitable for your CD drive, and one floppy connector which should work for your floppy drive just fine.
Peter
On Thu, 23 Jan 2014, Peter wrote:
On 01/23/2014 10:59 PM, Peter wrote:
On 01/23/2014 07:15 AM, Michael Hennebry wrote:
That said, it has fewer connectors. I had to leave off my CD and floppy drives.
I have yet to see a power supply that doesn't have connectors for these,
I just had trouble counting.
but you can get adapters and/or splitters for that if need be.
...in fact: http://raidmax.com/psu/rx_380k.html
it has four molex four pin connectors any one of which should be suitable for your CD drive, and one floppy connector which should work for your floppy drive just fine.
I needed the floppy connector for my video card.
On 01/24/2014 03:47 AM, Michael Hennebry wrote:
On Thu, 23 Jan 2014, Peter wrote:
it has four molex four pin connectors any one of which should be suitable for your CD drive, and one floppy connector which should work for your floppy drive just fine.
I needed the floppy connector for my video card.
Fair enough, you can get a four pin molex to floppy adapter and use that if you really care about that 1980's piece of technology.
Peter
On 1/23/2014 6:41 PM, Peter wrote:
On 01/24/2014 03:47 AM, Michael Hennebry wrote:
On Thu, 23 Jan 2014, Peter wrote:
it has four molex four pin connectors any one of which should be suitable for your CD drive, and one floppy connector which should work for your floppy drive just fine.
I needed the floppy connector for my video card.
Fair enough, you can get a four pin molex to floppy adapter and use that if you really care about that 1980's piece of technology.
Peter
Some have said it already, but to me it is rude to have a discussion about hardware problems on a software mailing list. Everyone who signed up for this would have signed up for CentOS.... OS being 'Operating System'. I don't know how many are on this list... thousands I would assume. Having a discussion about fixing computers belongs somewhere else.
Further, this list is archived in many locations. Off topic discussions degrade the quality of those archives when doing searches.
Best Regards, John Hinton
On 12/06/2013 06:57 PM, Michael Hennebry wrote:
I suspect the pincushiony thing between the video card and the big black Intel fan of being the heat sink for the CPU, but I do not know.
That case looks very dusty and 60C for an Intel CPU tells us that it is most likely overheating.
The big black Intel fan is the fan for the CPU heatsink - which is what it is physically mounted on. *Don't try to remove it.* Since you didn't recognize a CPU fan on sight you clearly have no background in disassembling and reassembling PCs and you will most likely damage the CPU before you are done. Failure to reattach the CPU cooling fan correctly (which involves cleaning off the old heatsink compound and applying new heatsink compound correctly) **will** cause CPU overheating and system problems and can damage the CPU.
I would start by gettting a can of compressed air, gently place a finger on the black Intel fan blades so it doesn't spin (spinning up a fan with air turns it into a generator pushing damaging voltage back into the motherboard - you don't want to do that) and blow the all the dust out of the heat sink for the CPU while moving the fan blades with your finger to allow access to the entire heatsink. Then boot the machine and verify that the CPU fan is in fact spinning.
Also blow the dust out of the power supply (the silver box at the top left) and off the fins of the video card.
Well just a note that sometimes CPU and other parts overload can cause a similar effect while newer software might offer better stability based on some sensors in the MB.
All The Bests, Eliezer
On 11/25/2013 07:45 AM, Michael Hennebry wrote:
CentOS 6.4 died on me again. Didn't leave any traces that I could find. The screen just suddenly went black. Couldn't switch to another virtual terminal. Pushing the reset button worked. Didn't have to power off this time.
-
Benjamin Franz -
Darr247 -
david -
Eliezer Croitoru -
Fernando Cassia -
Frank Cox -
Fred Smith -
John Hinton -
John R Pierce -
Karanbir Singh -
Les Mikesell -
Louis Lagendijk -
m.roth＠5-cent.us -
Mauricio Tavares -
Michael Hennebry -
Patrick Lists -
Peter -
SilverTip257 -
Steve Clark -
Thomas Dineen -
Warren Young