Hi,
I have built a new PC on which I've installed CentOS 5.5 64-bit (with updates) which after some hours of running suddenly either hard freeze or instant power off. The /var/log/messages log doesn't reveal anything of what has happened. The load has been either just idling or with moderate disc activities (transfers over 100GBit). Below is the hardware spec:
Motherboard: Gigabyte GA-H55N-USB3 CPU: Intel Core i5 650 RAM: CORSAIR 8 GB RAM HDD1: Intel X25-M 80GB HDD2-7: Samsung ECHOGREEN 500GB ST LABS SATA/300 4-PORT PCI-E PSU: Corsair CX 430W
I've also been running memtest86 for more than 24 hours without any errors.
I was on my way to return either the motherboard or the PSU when I thought of installing Fedora 14 and give it a try. The PC has now been running for over 48 hours without any glitch, with varying load.
I should also mention another problem that strangely disappeared with F14, and that was that the Samsung HDDs, which I have in a soft RAID6, now and then timed out (only one at a time). The message log showed a dump of NCQ queued commands, before it reconnected that disk. In F14 I haven't seen any errors at all in the log.
Could those problems only relate to C5, or do you think I also have hardware problems and only been lucky so far with F14?
Best Regards, Christian
I have built a new PC on which I've installed CentOS 5.5 64-bit (with updates) which after some hours of running suddenly either hard freeze or instant power off.
Can you check a setting in the bios - see if there's an option named: PCI Latency Timer
On Sun, Jan 23, 2011 at 00:14, compdoc compdoc@hotrodpc.com wrote:
I have built a new PC on which I've installed CentOS 5.5 64-bit (with updates) which after some hours of running suddenly either hard freeze or instant power off.
Can you check a setting in the bios - see if there's an option named: PCI Latency Timer
No, only some frequency adjustment of the PCI-Express bus.
Update: After some more days of running Fedora 14 those disk timeouts has started to appear. After a search on the net about similar NCQ problems I added this to /etc/rc.local yesterday, and so far I haven't seen any disk timeouts since then:
for D in sd{b,c,d,e,f,g} ; do echo 1 > /sys/block/$D/device/queue_depth done
Eventually I wish to install CentOS 6 and hopefully I will not have the same problem as with CentOS 5.5. In the mean time I will make a try with Scientific Linux 6 alpha or RHEL 6 beta2 and see what happens.
On Mon, Jan 24, 2011 at 22:48, cwlists cwlists@gmail.com wrote:
On Sun, Jan 23, 2011 at 00:14, compdoc compdoc@hotrodpc.com wrote:
I have built a new PC on which I've installed CentOS 5.5 64-bit (with updates) which after some hours of running suddenly either hard freeze or instant power off.
Can you check a setting in the bios - see if there's an option named: PCI Latency Timer
No, only some frequency adjustment of the PCI-Express bus.
Update: After some more days of running Fedora 14 those disk timeouts has started to appear. After a search on the net about similar NCQ problems I added this to /etc/rc.local yesterday, and so far I haven't seen any disk timeouts since then:
for D in sd{b,c,d,e,f,g} ; do echo 1 > /sys/block/$D/device/queue_depth done
Eventually I wish to install CentOS 6 and hopefully I will not have the same problem as with CentOS 5.5. In the mean time I will make a try with Scientific Linux 6 alpha or RHEL 6 beta2 and see what happens.
Update 2: Since the 25:th of January the PC has been running SL 6 (6rolling, dated 2011-01-21) without any problems at all (except the NCQ problem which has been solved by the earlier mentioned solution above).
I'm still a bit confused that an OS can make a PC to suddenly power off, but at least now I feel confident that I don't have to replace any part of the newly bought hardware if I run RHEL/CentOS/SL 6.
//Christian
-
compdoc -
cwlists