[zfs-discuss] resilvering zil/log mirror - expected behavior?
bill at bfccomputing.com
Fri Dec 16 12:08:30 EST 2011
I replaced a zil/log (10GB) mirror device today and was surprised by the resilver behavior. ZFS reported that it needed to resilver the entire used capacity of the zpool (1.82T), when I would have expected it to only resilver the log mirror.
I'm not sure what the expected behavior here is.
The actual process completed in 41 minutes, which I don't think is enough time to even read that much data from the hard drives (500GB might be possible - it's a linux md of 5400RPM 3TB disks with a LUKS wrapper). The L2ARC is only 30GB. Once it got rolling it was reporting resilver speeds of ~700M/s which only the SSD's can sustain:
270G scanned out of 1.82T at 706M/s, 0h38m to go
Perhaps there are references to that much data on hard disk on the ZIL? Ideally, I'd have expected to need only 10GB worth of resilver on the SSD's which would have just taken seconds, but probably I'm missing something about how ZFS works. I don't even know if it's actually a problem, and if it is, it's just one of efficiency.
# zpool attach backup scsi-SATA_INTEL_SSDSA2VP0CVLC1186007R020AGN-part2 scsi-SATA_INTEL_SSDSA2VP0CVLC118600MU020AGN-part2
# zpool status backup
scan: resilvered 0 in 0h41m with 0 errors on Fri Dec 16 10:33:57 2011
NAME STATE READ WRITE CKSUM
backup ONLINE 0 0 0
dm-0 ONLINE 0 0 0
mirror-1 ONLINE 0 0 0
scsi-SATA_INTEL_SSDSA2VP0CVLC1186007R020AGN-part2 ONLINE 0 0 0
scsi-SATA_INTEL_SSDSA2VP0CVLC118600MU020AGN-part2 ONLINE 0 0 0
scsi-SATA_TS64GSSD25S-M_20110211405406066105-part2 ONLINE 0 0 0
errors: No known data errors
Bill McGonigle, Owner
BFC Computing, LLC
Email, IM, VOIP: bill at bfccomputing.com
Social networks: bill_mcgonigle/bill.mcgonigle
More information about the zfs-discuss