I have an X4500 fileserver (NFS, Samba) running OpenSolaris 2008.05 pkg
upgraded to snv_91 with ~3200 filesystems (and ~27429 datasets, including
snapshots).
I've been encountering some pretty big slow-downs on this system when running
certain zfs commands. The one causing me the most pain at the moment is
setting the "sharenfs" property on a filesystem takes a little under 7 minutes
per operation.
Tracing it, I see that it's making a huge number of ioctl calls:
$ truss -f -c zfs set sharenfs=on zpool/test
syscall seconds calls errors
_exit .000 1
read .150 8618
open .166 6826 2
close .049 6839
brk .022 1246
lseek .000 3
getpid .039 7357
sysi86 .000 1
ioctl 32.834 6008913 10239
execve .000 1
fcntl .000 18
openat .000 1
getppriv .000 1
getprivimplinfo .000 1
issetugid .000 4
sigaction .000 1
sigfillset .000 1
getcontext .000 1
setustack .000 1
mmap .000 74
munmap .000 29
xstat .003 43 19
getrlimit .000 1
memcntl .000 15
sysconfig .000 5
lwp_sigmask .000 2
lwp_private .000 1
llseek .089 13688
door_info .000 14
door_call .086 7130
schedctl .000 1
resolvepath .000 18
getdents64 .000 2
stat64 .000 3
fstat64 .112 20377
zone_getattr .000 1
zone_lookup .000 3
-------- ------ ----
sys totals: 33.556 6081241 10260
usr time: 41.354
elapsed: 386.130
Looking at exactly which ioctls get called:
5991763 MNTIOC_GETMNTENT
6782 TCGETA
6892 ZFS_IOC_DATASET_LIST_NEXT
3466 ZFS_IOC_OBJSET_STATS
2 ZFS_IOC_POOL_CONFIGS
1 ZFS_IOC_SET_PROP
1 ZFS_IOC_SHARE
8 ZFS_IOC_SNAPSHOT_LIST_NEXT
Is anyone else experiencing this, or does anyone have any ideas on how to solve
it? Is this solved in more recently snv releases? (And are there changelogs
posted somewhere?)
Thanks!
-Laen
--
This message posted from opensolaris.org
_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
