It seems I can't smarm monitor disks on RAID controllers? I tried on several machines with two different controllers and on all I get this when starting up smartd:
Jul 27 14:36:43 c1 smartd[5944]: Opened configuration file /etc/smartd.conf Jul 27 14:36:43 c1 smartd[5944]: Configuration file /etc/smartd.conf parsed. Jul 27 14:36:43 c1 smartd[5944]: Device: /dev/sda, opened Jul 27 14:36:43 c1 smartd[5944]: Device: /dev/sda, Bad IEC (SMART) mode page, err=4, skip device Jul 27 14:36:43 c1 smartd[5944]: Unable to register SCSI device /dev/sda at line 32 of file /etc/smartd.conf Jul 27 14:36:43 c1 smartd[5944]: Device /dev/sda not available
/dev/sda is the virtual disk as it appears to CentOS as I can access it with hdparm. Do I need to use another device for the RAID array (which?) or is it impossible to smart monitor "thru" a RAID controller?
Kai
On Sun, Jul 27, 2008 at 8:51 AM, Kai Schaetzl maillists@conactive.com wrote:
It seems I can't smarm monitor disks on RAID controllers? I tried on several machines with two different controllers and on all I get this when starting up smartd:
/dev/sda is the virtual disk as it appears to CentOS as I can access it with hdparm. Do I need to use another device for the RAID array (which?) or is it impossible to smart monitor "thru" a RAID controller?
This depends on the raid controller. 3ware and a few others will pass info, though you have to tell smart about them in the config. Many hardware raid controllers will not pass smart info, as it's not just one drive you're watching, and the vendor has their own solution for that.
Jim Perrin wrote on Sun, 27 Jul 2008 09:07:19 -0400:
This depends on the raid controller.
Ah, I see. I listed them in the reply to Rainer.
Kai
Am 27.07.2008 um 14:51 schrieb Kai Schaetzl:
It seems I can't smarm monitor disks on RAID controllers? I tried on several machines with two different controllers and on all I get this when starting up smartd:
Jul 27 14:36:43 c1 smartd[5944]: Opened configuration file /etc/smartd.conf Jul 27 14:36:43 c1 smartd[5944]: Configuration file /etc/smartd.conf parsed. Jul 27 14:36:43 c1 smartd[5944]: Device: /dev/sda, opened Jul 27 14:36:43 c1 smartd[5944]: Device: /dev/sda, Bad IEC (SMART) mode page, err=4, skip device Jul 27 14:36:43 c1 smartd[5944]: Unable to register SCSI device /dev/ sda at line 32 of file /etc/smartd.conf Jul 27 14:36:43 c1 smartd[5944]: Device /dev/sda not available
/dev/sda is the virtual disk as it appears to CentOS as I can access it with hdparm. Do I need to use another device for the RAID array (which?) or is it impossible to smart monitor "thru" a RAID controller?
What controller would that be?
Rainer
Rainer Duffner wrote on Sun, 27 Jul 2008 15:39:34 +0200:
What controller would that be?
It's Dell SAS 6iR and HP "8 Port HBA Host Controller" (as they call it). Both are LSI 1068-based, not sure if exactly the same chip.
Kai
/dev/sda is the virtual disk as it appears to CentOS as I can access it with hdparm. Do I need to use another device for the RAID array (which?) or is it impossible to smart monitor "thru" a RAID controller?
Kai
Kai
Shouldn't there be another "layer" ?
Is this a real hardware raid controller?
I do not remember the specifics, yet when we did this on HP proliant boxes with hardware raid, we had to play with the config files and setup for the raid *drivers* so to speak
On old boxes with raid5 we would be monitoring something like
/dev/ida/c0d0
And on newer boxes with raid1 or raid5 we would be monitoring something like
/dev/cciss/c0d0
- rh
Robert - elists wrote on Sun, 27 Jul 2008 08:35:24 -0700:
/dev/ida/c0d0 /dev/cciss/c0d0
I looked around, but none of these :-( /dev/disk points to pci ids and then to /dev/sda etc. It's a PCI-based controller that HP calls "HP 8 Port HBA Controller" or so. In a DL140 G3.
Kai
Am 27.07.2008 um 20:31 schrieb Kai Schaetzl:
Robert - elists wrote on Sun, 27 Jul 2008 08:35:24 -0700:
/dev/ida/c0d0 /dev/cciss/c0d0
I looked around, but none of these :-( /dev/disk points to pci ids and then to /dev/sda etc. It's a PCI-based controller that HP calls "HP 8 Port HBA Controller" or so. In a DL140 G3.
For HP, you should get the data if you install the Insight Manager agents. For Dell - I don't know.
cu, Rainer
Rainer Duffner wrote on Mon, 28 Jul 2008 00:22:03 +0200:
For HP, you should get the data if you install the Insight Manager agents.
I didn't install any software from HP on the HP machine. After half-an- hour searching on the site I finally found an "hpasm" package that seems to be it. They make it really hard to find it as it doesn't appear when searching for "drivers and software" for the server. Unfortunately, that didn't change anything. hpasmd is running, but hpasmcli just hangs when I try it out. There was an hp-OpenIPMI packaged mentioned, I'll search for that - tomorrow. Thanks for the hint, anyway.
Kai
On Mon, Jul 28, 2008 at 11:31 AM, Kai Schaetzl maillists@conactive.com wrote:
I didn't install any software from HP on the HP machine. After half-an- hour searching on the site I finally found an "hpasm" package that seems to be it. They make it really hard to find it as it doesn't appear when searching for "drivers and software" for the server. Unfortunately, that didn't change anything. hpasmd is running, but hpasmcli just hangs when I try it out. There was an hp-OpenIPMI packaged mentioned, I'll search for that - tomorrow. Thanks for the hint, anyway.
I'm just guessing here, but it seems like "hpasm" is just there to make you feel good on their machines, especially if they make it hard for you to find it. Or maybe it makes the machine feel good....
SCNR
mhr
Mhr wrote on Mon, 28 Jul 2008 11:41:17 -0700:
Or maybe it makes the machine feel good....
Funny that you say that. Believe it or not, but after I found that hpasm didn't provide any useful for me (at least at the moment) and I shut down the daemon (with all of its agents) the core temperatures in sensors dropped within a few seconds by 4 degrees.
Kai
Am 29.07.2008 um 00:31 schrieb Kai Schaetzl:
Mhr wrote on Mon, 28 Jul 2008 11:41:17 -0700:
Or maybe it makes the machine feel good....
Funny that you say that. Believe it or not, but after I found that hpasm didn't provide any useful for me (at least at the moment) and I shut down the daemon (with all of its agents) the core temperatures in sensors dropped within a few seconds by 4 degrees.
It's know for consuming some CPU. But taking a quick look on the support site (HP.com -> Support/Drivers-
Servers->Proliant->DL>140>G3), I see that they consider your DL140 a
bit too small to support with all tools.
Normally, something like the HP Array Diagnostics utility would do the job, but alas: "HP ProLiant 100-series customers: Because the HP System Management Homepage software is not supported for use with HP ProLiant 100-series servers, the Array Diagnostics Utility for Linux cannot be installed."
Rainer
Rainer Duffner wrote on Tue, 29 Jul 2008 00:46:58 +0200:
It's know for consuming some CPU.
A-h-a.
But taking a quick look on the support site (HP.com -> Support/Drivers-
Servers->Proliant->DL>140>G3),
tried that, I just reach the same "arena" "Download drivers and software - HP ProLiant DL140 G3 Server series" that I know. I have a shortcut for that here ;-)
I see that they consider your DL140 a bit too small to support with all tools.
Normally, something like the HP Array Diagnostics utility would do the job, but alas: "HP ProLiant 100-series customers: Because the HP System Management Homepage software is not supported for use with HP ProLiant 100-series servers, the Array Diagnostics Utility for Linux cannot be installed."
How did you get that? I can't find a statement like that anywhere. Not with release notes, installations instructions, not even here: ftp://ftp.hp.com/pub/softlib2/software1/pubsw- linux/p1174222445/v47112/hpadu-8.10-3.noarch.txt
Anyway, the point about this is that the DL140 by itself does not have "real" RAID controllers, they seem to be fake RAID ones. That's why the software is not supposed to work on them. However, it can also be shipped with a "real" RAID PCI card. That is what I have, they call it "HP 4/8 Port HBA Controller" or something. I don't know why but they refuse to give it a "real" name. However, support told me that the "HP Array" utilities are not supposed to work on that either, they are only supposed to work on internal controllers (and the readme doesn't list this controller, either). So, which way you look at it, there's a black hole ;-) I'm gonna live without smart for that machine.
Kai
On Mon, Jul 28, 2008 at 1:31 PM, Kai Schaetzl maillists@conactive.com wrote:
Rainer Duffner wrote on Mon, 28 Jul 2008 00:22:03 +0200:
For HP, you should get the data if you install the Insight Manager agents.
I didn't install any software from HP on the HP machine. After half-an- hour searching on the site I finally found an "hpasm" package that seems to be it. They make it really hard to find it as it doesn't appear when searching for "drivers and software" for the server. Unfortunately, that didn't change anything. hpasmd is running, but hpasmcli just hangs when I try it out. There was an hp-OpenIPMI packaged mentioned, I'll search for that - tomorrow. Thanks for the hint, anyway.
IMHO, the HP Support site is *very* difficult to navigate, if one is not there frequently; in comparison to the Dell Support site. Once you get to the right place, they probably have what you need. I won a Compaq Evo D300v some years ago in a raffle. The BIOS in it seems far advanced, compared to our 4 Dell Dimension boxes, because it is for Enterprise use.
Kai Schaetzl wrote: ...
Unfortunately, that didn't change anything. hpasmd is running, but hpasmcli just hangs when I try it out.
I run CentOS 5 on our old ML 370 G3 machines.
BigBrother is watching the machines, using /sbin/hplog from hpasm to query the state of the fans, powersupplies and temperatures.
The state of the smart array RAID is monitored with array-info: https://sourceforge.net/projects/array-info/
Mogens
Mogens Kjaer wrote on Tue, 29 Jul 2008 12:14:16 +0200:
BigBrother is watching the machines, using /sbin/hplog from hpasm to query the state of the fans, powersupplies and temperatures.
Thanks for the tip. Unfortunately, that hangs like hpasmcli is hanging. I may be missing something (kernel module, drivers?), but I don't know which. I didn't install any HP drivers as I prefer to use the CentOS drivers if they work. This way I can avoid any trouble when upgrading the kernel. Maybe I'm missing something this way? So far I only installed that extra hpasm package and the hp-OpenIPMI. I can query sensors on the Dells with ipmitool, but this doesn't work on the HP, it can't find an ipmi device. A configure run of hpasm shows me "This server does not contain an hp Advanced Server Management Device.". That's probably the reason for the hang. However, there's a Lights Out BMC built-in and I can login to it. I didn't acquire licenses for virtual KVM and virtual media, but I think this shouldn't matter in this case. Hah, got it. "service ipmi start" loads all ipmi modules (there was at least one missing before) and I get a read-out with ipmitool. hpasmcli and hplog still don't work. I think I will just forget about hpasm.
The state of the smart array RAID is monitored with array-info: https://sourceforge.net/projects/array-info
won't compile without docbook. But I think it won't work with the controller, anyway. I can get a reading with "mpt-status -i 1" for the RAID status (on both HP and Dell machines), but this doesn't include SMART (SMART as in Self-Monitoring, Analysis, and Reporting Technology, I think the "smart" in Smart Array refers to something different?).
Kai
Kai Schaetzl wrote: ...
won't compile without docbook. But I think it won't work with the controller, anyway. I can get a reading with "mpt-status -i 1" for the RAID status (on both HP and Dell machines), but this doesn't include SMART (SMART as in Self-Monitoring, Analysis, and Reporting Technology, I think the "smart" in Smart Array refers to something different?).
Why are you interested in the SMART information?
The RAID controller should watch the SMART data and issue a warning if a drive is about to fail, possibly kicking in a hot spare.
Mogens
Mogens Kjaer wrote on Wed, 30 Jul 2008 08:30:11 +0200:
Why are you interested in the SMART information?
Curiousity ;-) I'm not familiar with RAID controllers, so I don't know yet what they might do or not. Thanks for the info!
Kai
/dev/sda is the virtual disk as it appears to CentOS as I can access it with hdparm. Do I need to use another device for the RAID array (which?) or is it impossible to smart monitor "thru" a RAID controller?
You probably will want to install the "HP Proliant Support Pack" as it will include the hpasm command line tools as well as a handy HP System Management home page. You can set it up to send you email alerts or send snmp traps to a central server running the HP SIM software any time one of your hard drives fails. We use HP SIM in production to monitor both our Dell and HP servers, and it works great. I get an alert any time a RAID card or hard drive has issues.
-
Jim Perrin -
Kai Schaetzl -
Lanny Marcus -
MHR -
Mogens Kjaer -
Rainer Duffner -
Robert - elists -
Sean Carolan