when ZFS is used as ZFS's primitives like osd_write()
and osd_declare_write() are still few times slower
compared to ldiskfs and this 144b creating wide-striped
files which causes lots of tiny writes to the last-used
file.
Test-Parameters: trivial testlist=recovery-small
Test-Parameters: testlist=recovery-small fstype=zfs env=ONLY=144b,ONLY_REPEAT=100
Signed-off-by: Alex Zhuravlev <bzzz@whamcloud.com>
Change-Id: Ib023c7c71a3bb486f9f9908e4cc03cc6e53ace7a
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/49693
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Arshad Hussain <arshad.hussain@aeoncomputing.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
kill -9 $pid >/dev/null 2>&1
done
- before=$(date +%s)
+ before=$SECONDS
fail mds1
- after=$(date +%s)
+ after=$SECONDS
# here we measure MDT stop + MDT start time. For error case MDT stop takes
# about obd_timeout-60 (240) seconds. Without error - less than 30s.
# MDT start takes different time depends on a configuration, let's check
done
fail ost1
- sleep 60
+
+ # 60s is not enought with ZFS which is still significatnly
+ # slower on some operations like record writing
+ local stime=60
+ [[ "$mds1_FSTYPE" != "zfs" ]] || stime=120
+ sleep $stime
for pid in $pids; do
ps -p $pid > /dev/null