I'm testing out iSCSI at the moment, creating a ZFS volume out of iSCSI devices. I've found that if I disconnect an iSCSI device, file I/O and the "zpool status" command will hang for a little over 5 minutes before the system finally realises something is wrong.
Is there any way I can change this? Ideally I'd like it to fail within 5-10 seconds. Despite a few days of searching, the only thing I can find vaguely relvant is this: http://mail.opensolaris.org/pipermail/storage-discuss/2006-May/001646.html. Towards the bottom of that post, it mentions a scsi_pkt timeout of 60 seconds, and states that the disk driver will recover from outages of up to 90 seconds, but will fail if you leave the interface down long enough. It also says that if you have fastfail enabled, for clustering etc, the I/O on the path will fail right away. It sounds like fast-fail might be what I need, but I can't find anything about what it is, or how to configure it. Can anybody help? This message posted from opensolaris.org _______________________________________________ storage-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/storage-discuss
