HAST + ZFS + NFS + CARP

[Borja Marcos]

> On 11 Aug 2016, at 11:39, Ben RUBSON <ben.rubson at gmail.com> wrote:
> 
> 
>> On 11 Aug 2016, at 11:24, Borja Marcos <borjam at sarenet.es> wrote:
>> 
>> Although, frankly,
>> ZFS is extremely resilient. One of mine even survived a SAS HBA problem that caused some
>> silent corruption.
> 
> Any link to this issue Borja ?
> Thank you !

It wasn’t a FreeBSD or ZFS bug, but a defective part (a HBA). Once in a while we saw some errors in /var/log/messages
and zfs scrub revealed some corruption that ZFS fixed without issues. Determining the cause wasn’t easy (at first it looked
like a defective backplane) and IBM, who are no longer welcome here thanks to their totally fabulous support and warranty
policy, didn’t help much. So we took the system offline, using the replicated server instead, and it took some time doing tests
(during which we caused more silent corrption which ZFS fixed without problems) to determine that it was indeed the HBA.

Finally we replaced the HBA and the system is back at work. But not a single bit was lost.





Borja.