On Thu, Apr 5, 2012 at 4:48 AM, Weber, Markus <f...@de.kpn-eurorings.net> wrote:

> Even though it's not directly ZFS related, I've seen some similar discussion
> on this list and maybe someone has "the final" answer to this problem, as most
> tips and "these things could help" I have found so far have not fully solved
> the problem.
> We are struggling with the behaviour of the combination LSI 3081E-R and SATA
> disks behind an expander.
> One disk behind the expander is known to be bad. DDing from that disk causes
> I/O to other (good) disks to fail soon (Solaris) or later (Linux), but for 
> sure
> it will fail and make the system unusable.


    We have five J4400 loaded with SATA drives connected to two dual
port 1068E based controllers. A fully supported configuration as of
when we bought it. We also two J4400 loaded with SATA drives behind a
single dual port 1068E based controller. We also have three instances
of a single J4400 behind a dual port 1068E controller. In all cases
when I say "dual port" I mean dual external SAS connector, each with 4
channels, so a total of 8 channels per controller. All J4400 are dual
attached and we are running Solaris 10U9 with multi-pathing enabled.

    I have not seen any odd issues with the five J4400 configuration
since we went production. In pre-production testing we found a bug in
the MPT driver that would cause a failed dead drive go undetected for
_hours_ while zfs blindly trusted the FMD layer and kept issuing I/O
requests and waiting for responses that were never coming back. This
was fixed in an IDR (which we are running) but has been fully
integrated in 10U10.

    I have seen odd behavior of the single J4400 configurations when a
drive fails. I have not been able to really qualify the problem, just
very slow I/O and no logs to point at anything other than the single
failed drive. Sometimes reseating the failed drive will make it come
back to life, sometimes for a short while sometimes (apparently)

    I have not seen any odd behavior due to the J4400 in the two J4400
configuration (we have had other issues with this system, but they
were not related to the J4400).

    No data has been lost due to any of the failures or outages. Thank you ZFS.

Paul Kraus
-> Senior Systems Architect, Garnet River ( http://www.garnetriver.com/ )
-> Sound Coordinator, Schenectady Light Opera Company (
http://www.sloctheater.org/ )
-> Technical Advisor, Troy Civic Theatre Company
-> Technical Advisor, RPI Players
zfs-discuss mailing list

Reply via email to