zfs stuck, cannot do any I/O, processes in Disk Wait
Adam McDougall
mcdouga9 at egr.msu.edu
Mon Oct 29 11:49:00 PDT 2007
I think I have had this happen at least once before, but someone else
rebooted the system before I could see it. I have a server with a
number of zfs filesystems mounted from a raidz, but it won't transfer
any data. I'm not sure why its stuck. It is running 7.0-PRERELEASE
Wed Oct 17 and I'm pretty sure it is WITHOUT vm_kern.c.2.patch. The
system is amd64 and I have not seen a kmem panic since I raised kmem
to 1.5G.
I logged in to scp a file off of zfs, was able to ls -l to see the file
but the scp hung before transferring any bytes. Now I cannot do a ls -l
in that directory, /z. I noticed several days worth of rsync processes
stuck in disk wait, it must have been in this state for several days.
I have no urgent need to reboot this system, its more important to try
to get a permanent fix. Please let me know what other information I
can provide.
10:34PM up 10 days, 12:37, 3 users, load averages: 0.00, 0.00, 0.00
Stuck rsync processes (started from cron):
1:01AM
4:00AM
5:00AM
Fri01AM
Fri04AM
Fri05AM
Mon05AM
Sat01AM
Sat04AM
Sat05AM
Thu01AM
Thu04AM
Thu05AM
Tue01AM
Tue04AM
Tue05AM
Wed01AM
Wed04AM
Wed05AM
# more /boot/loader.conf
vm.kmem_size=1610612736
vm.kmem_size_max=1610612736
# sysctl -a | grep vnodes
kern.maxvnodes: 100000
kern.minvnodes: 25000
vfs.freevnodes: 25000
vfs.wantfreevnodes: 25000
vfs.numvnodes: 49996
No errors in dmesg, and I can dd from the drives in the raidz1 fine.
z/backups 101508480 0 101508480 0% /backups
z/backups/a 101508480 0 101508480 0% /backups/a
z/backups/b 149992448 48483968 101508480 32% /backups/b
z/backups/c 219571968 118063488 101508480 54% /backups/c
z/backups/d 105923968 4415488 101508480 4% /backups/d
z/data 199868032 98359552 101508480 49% /data
z 103982976 2474496 101508480 2% /z
z/data4 206146688 104638208 101508480 51% /z/data4
z/mysqldb 102015488 507008 101508480 0% /z/mysqldb
# zpool list
NAME SIZE USED AVAIL CAP HEALTH ALTROOT
z 696G 540G 156G 77% ONLINE -
# zpool status
pool: z
state: ONLINE
scrub: none requested
config:
NAME STATE READ WRITE CKSUM
z ONLINE 0 0 0
raidz1 ONLINE 0 0 0
ad4 ONLINE 0 0 0
ad6 ONLINE 0 0 0
ad8 ONLINE 0 0 0
errors: No known data errors
More information about the freebsd-current
mailing list