I want to monitor the hardware health, not have my health ruined mine through stress

July 18, 2010 11:02 PM

What rackmountable or blade hardware vendor has the best support for hardware monitoring x86 Linux?

We currently use HP servers and blades and the Proliant Support Pack software HP has us install to provide hardware health monitoring is nothing but trouble. broken installs, an ever moving target, random kernel panics... Are the pastures really greener with Dell or IBM's offering? We are running RHEL5 and a few RHEL4 systems. Most of are problems are related to HBA drivers and SAN booting, but that's just the tip of the iceberg when it comes to the headaches I've received from PSP. I'm ready to start evaluating other hardware vendors, but I'm curious if this will just introduce another set of crappy proprietary drivers.

posted by jrishel to technology (7 answers total)

Well it depends how you will monitor the monitoring? Are you using SNMP?

IBM's agent is pretty good. Haven't seen any crashes caused by it.

Fujitsu's agent is also pretty good now (up until about a year ago it was a pain to install). One of the daemons sometimes likes cpu but it hasn't caused any problems.
posted by devnull at 1:19 AM on July 19, 2010

This sounds like a job for ... A VALUE-ADDED RESELLER! Seriously, the only way to navigate vendor baloney while still being productive is to enlist a vendor-neutral VAR. These guys have the ability to sell you anything, and if they're good, their cut is about the same no matter what vendor you purchase. A really good one will come out, assess your situation, and make recommendations based on that. They usually have engineers who will know what they're talking about.

If you can find a good VAR, hold on to them, because in my experience the ones I work with made my life so much easier.

That said, I never had any driver issues with Dell blades. I don't remember the model, but they were top-of-the-line 1.5 years ago. I set up an HPC cluster, and no one's touched it since.
posted by Geckwoistmeinauto at 4:37 AM on July 19, 2010

We're using Openview (snmp and agents) to monitor, but that may change soon too. As an organization, we're less that happy with HP lately.
posted by jrishel at 4:46 AM on July 19, 2010

I've found Dell's server support in the Linux space to be pretty good. They've got a YUM repository you can add to all your servers that makes installing their stuff real easy, and there's a wiki for it with decent documentation. You can check it out here:

http://linux.dell.com/wiki/index.php/Repository/hardware
posted by word_virus at 7:42 AM on July 19, 2010

I'm biased although I don't work in the hardware space. Have you worked with your support contacts at HP? I've seen the internal mailing lists and I see a lot of people working hard to make their customers very happy.
posted by mmascolino at 7:56 AM on July 19, 2010

HP's support does work very hard fix our issues... but we seem to keep having issues...
posted by jrishel at 8:57 AM on July 19, 2010

I don't know if this is helpful advice, but if you don't want to worry about hardware failures and crap drivers, abstract them out by virtualizing. booting to a cloned SD card with SAN-attached storage is paradise. From worst to best, we worked through Hyper-V, Xen, KVM, and VMWare.

Now, our hardware monitoring is done via nagios.

Otherwise, IBM pledged a billion dollars towards linux and I think it shows.
posted by overbo at 12:44 AM on August 6, 2010