I'm using snv_111 to host iSCSI for my backups. This went fine until I enabled compression on the volume. About halfway through a backup (~250gb done), Solaris loses its network connection with no errors logged (/var/adm/messages and /var/log/* with no entries for an hour preceding). After reformatting the iSCSI volume (from Windows) and starting from scratch to use compression, it takes about 10gb to trigger this. I've tried several switches and routers, and the same always happens - the Solaris system drops off the network for about 15 minutes.
I tried to do a 'zfs destroy' on the volume, and it's been hung, along with all other zfs commands for about 3 hours now. The volume only ever contained 550gb total of data. This is a 700gb volume on two mirrored 1tb drives. Looking back through the mailing list, I see the zfs destroy hanging problem, but it looked like people believed that to be more related to dedup. -- This message posted from opensolaris.org