PDA

View Full Version : Kernel Panic on Proliant DL360P



jqueiroz
20-Oct-2014, 19:54
Hi,

We have a bunch of HP Proliant DL360P servers, running SLES 11 SP3 (and OES 11 SP2). These machines have a internal RAID controller (B120i), which uses a driver from HP named "hpvsa".

We also use a few Lenovo ix4-300d storages linked to these servers, working as an iSCSI source, so we can have a Novell Cluster on our branch offices.

We've seen that when, for any reason, the servers loose contact with the iSCSI host (the storage), system throws a kernel panic, related to the hpvsa driver.

Does anyone have seen this before? This situation is causing a lot of trouble, specially in one of the branches (which, for some reason, have its' switches reset nightly --- so I have to reset both servers almost every morning).

jmozdzen
21-Oct-2014, 12:22
Hi jqueiroz,

> We've seen that [...] system throws a kernel panic, related to the hpvsa driver.

have you seen this article from HP? http://h20565.www2.hp.com/portal/site/hpsc/template.PAGE/public/kb/docDisplay?docId=c03732112&ac.admitted=1413890430068.876444892.492883150

Regards,
Jens

jqueiroz
21-Oct-2014, 18:33
Yes, I saw... but I don't think is the case, as this feature seems to be AMD-related, and our servers are Intel-based.
And we have 8 servers intalled, and the problem only occurs on 4 of them.

jmozdzen
22-Oct-2014, 13:24
Hi jqueiroz,

Yes, I saw... but I don't think is the case, as this feature seems to be AMD-related, and our servers are Intel-based.
And we have 8 servers intalled, and the problem only occurs on 4 of them.

as I cannot exactly tell the source of the difficulties, I'm mostly guessing here - but IOMMU, as a general term, might also apply to Intel hardware (there it's called VT-x, see http://en.wikipedia.org/wiki/IOMMU). So if you're in a position to test, I'd recommend booting with "iommu=soft" to see if it makes any difference.

A second approach would be to search for differences between the four affected and the four non-affected servers. But as many different aspects might influence the problem, that easily can become rather tedious.

Regards,
Jens

jqueiroz
22-Oct-2014, 20:11
I found an update from HP to the HPVSA KMP, and I'll try to apply it. If the problem persists, I'll try the "iommu=soft".

Thanks a lot from your attention.