Is it still the case that there is a kernel panic if the device(s) with the ZFS pool die? I was thinking to attach some cheap SATA disks to a system to use for nearline storage. Then I could use ZFS send/recv on the local system (without ssh) to keep backups of the stuff in the main pool. The main pool is replicated across 2 arrays on the SAN and we have multipathing and it''s quite robust. However I don''t want my mail server going DOWN if the non-replicated pool goes down. I tested with a fully-patched 10u4 system and this panic still occurs. I don''t have at this moment a Nevada system to test because that system got pulled apart and it''ll be a few days before I can put something together again. Anyone have the current scoop? Thanks! This message posted from opensolaris.org
On Mon, 2008-02-25 at 11:48 -0800, Vincent Fox wrote:> Is it still the case that there is a kernel panic if the device(s) with the ZFS pool die? > > I was thinking to attach some cheap SATA disks to a system to use for nearline storage. Then I could use ZFS send/recv on the local system (without ssh) to keep backups of the stuff in the main pool. The main pool is replicated across 2 arrays on the SAN and we have multipathing and it''s quite robust. > > However I don''t want my mail server going DOWN if the non-replicated pool goes down. > > I tested with a fully-patched 10u4 system and this panic still occurs. I don''t have at this moment a Nevada system to test because that system got pulled apart and it''ll be a few days before I can put something together again. Anyone have the current scoop? > > Thanks! >PSARC 2007/567 added the "failmode" option to the zpool(1) command to specify what happens when a pool fails. This was integrated in Nevada b77, it probably won''t be available in S10 until the next update. -Albert
Having just installed Solaris 10 U5 I was kind of hoping that this was incorporated. This is a showstopper as far as using ZFS in production. This is because all production is based on EMC storage with either a backend RAID1 or RAID5. This is not an issue when systems have dual paths to storage but on dev systems which are single path this is a show stopper. Are there any ETA''s for when this "bug" is likely to be fixed as a patch. Veltror In order to have a future, live your past. This message posted from opensolaris.org