[OpenIndiana-discuss] Slow zfs list on oi_148 compared to snv_111b
Shawn Moore
smmoore at gmail.com
Thu Jun 16 20:05:37 UTC 2011
I have several(5) ZFS machines that have been built to serve out over
NFS. Most are OpenSolaris snv_111b, but we have two that are oi_148.
Between these we have ~ 200 volumes and 221000 snapshots (as of this
writing). On all of the oi_148's it takes a really really long time to
list the snapshots, but on the snv_111b's it's really quick. It does
not appear to be related to the amount of data or hardware as the
oi_148's have 144GB ram, 2x6 CPU's and even mirror'd SSD ZIL's. The
snv_111b's have 16GB ram, 2x4 CPU's and no SSD. Any help or
troubleshooting advice is greatly appreciated as I would like to move
to OI for several reasons. Below is my findings.
nas0 (oi_148):
NAME SIZE ALLOC FREE CAP DEDUP HEALTH ALTROOT
nas0 3.97T 1.88T 2.09T 47% 1.02x ONLINE -
# time zfs list -H -o name -r nas0 | wc -l
22
real 0m0.926s
user 0m0.009s
sys 0m0.022s
# time zfs list -H -o name -t all -r nas0 | wc -l
8949
real 25m26.583s
user 0m1.475s
sys 0m14.229s
nas1 (oi_148):
NAME SIZE ALLOC FREE CAP DEDUP HEALTH ALTROOT
nas1 1.98T 1.00T 1003G 50% 1.06x ONLINE -
# time zfs list -H -o name -r nas1 | wc -l
25
real 0m0.015s
user 0m0.005s
sys 0m0.013s
# time zfs list -H -o name -t all -r nas1 | wc -l
10710
real 4m7.599s
user 0m0.943s
sys 0m3.816s
nas3 (snv_111b):
NAME SIZE USED AVAIL CAP HEALTH ALTROOT
nas3 11.9T 9.51T 2.42T 79% ONLINE -
# time zfs list -H -o name -r nas3 | wc -l
67
real 0m0.347s
user 0m0.008s
sys 0m0.015s
# time zfs list -H -o name -t all -r nas3 | wc -l
98445
real 0m17.147s
user 0m6.160s
sys 0m10.835s
nas9 (snv_111b):
NAME SIZE USED AVAIL CAP HEALTH ALTROOT
nas9 5.97T 4.19T 1.77T 70% ONLINE -
# time zfs list -H -o name -r nas9 | wc -l
29
real 0m0.232s
user 0m0.005s
sys 0m0.012s
# time zfs list -H -o name -t all -r nas9 | wc -l
40846
real 0m7.138s
user 0m2.504s
sys 0m4.351s
More information about the OpenIndiana-discuss
mailing list