topic Re: RAID1+0 Issue in Disk Enclosures

RAID1+0 Issue

David Hill_7 — Fri, 17 Sep 2004 05:57:12 GMT

We have a HP Proliant DL580G1 (BIOS 386P20 05-Apr-2001) with 3 * Smart Array 5300 Controllers (Firmware 1.76). The O/S is Windows NT4 (SP6A). The array that had the issue is setup as follows: RAID1+0 Array (128K Stripe) 121GB, 14* 18.2GB disks with 50% Read, 50% Write Cache. We had an incident where one disk failed in the array, but the entire volume became inaccessible until the faulty disk was replaced. Has anyone seen similar incidents where fault tolerant arrays have not proven to be fault tolerant?

Re: RAID1+0 Issue

Uwe Zessin — Fri, 17 Sep 2004 09:44:50 GMT

Hello David,
if all disk drives live on a single SCSI bus, then a faulty drive can block the whole bus which results in a total block of the logical disk. I have already seen it (not on a SmartArray, but on other storage arrays).

Re: RAID1+0 Issue

Todd McDaniel_1 — Fri, 17 Sep 2004 12:03:01 GMT

I have seen something similar where when I get Lbolt errors down one disk that the bus can be distrupted and make disks appear to have write errors...

Or experience multiple time out errors which result in additional lbolts in most cases.

I havent seen any case where those disks were inaccessable though.

Re: RAID1+0 Issue

Uwe Zessin — Fri, 17 Sep 2004 13:49:34 GMT

What does 'Lbolt' mean, by the way? I have seen it mentioned in the HP-UX groups before, but forgot to ask.

Re: RAID1+0 Issue

Michael Schulte zur Sur — Fri, 17 Sep 2004 19:27:46 GMT

Uwe,

seen from an answer from ACS:
LBOLT indicates the number of clock ticks since the box was last booted

greetings,

Michael