zfs discuss - Jul 2007 - zfs list hangs if zfs send is killed (leaving zfs receive process)

I was in the process of doing a large zfs send | zfs receive when I decided that
I wanted to terminate the the zfs send process. I killed it, but the zfs receive
doesn''t want to die... In the meantime my zfs list command just hangs.

Here is the tail end of the truss output from a "truss zfs list":

ioctl(3, ZFS_IOC_OBJSET_STATS, 0x08043484)      = 0
ioctl(3, ZFS_IOC_SNAPSHOT_LIST_NEXT, 0x08045788) Err#3 ESRCH
ioctl(3, ZFS_IOC_DATASET_LIST_NEXT, 0x08046950) = 0
ioctl(3, ZFS_IOC_OBJSET_STATS, 0x0804464C)      = 0
ioctl(3, ZFS_IOC_DATASET_LIST_NEXT, 0x08045788) Err#3 ESRCH
ioctl(3, ZFS_IOC_SNAPSHOT_LIST_NEXT, 0x08045788) = 0
ioctl(3, ZFS_IOC_OBJSET_STATS, 0x08043484)      = 0
ioctl(3, ZFS_IOC_SNAPSHOT_LIST_NEXT, 0x08045788) Err#3 ESRCH
ioctl(3, ZFS_IOC_DATASET_LIST_NEXT, 0x08046950) = 0
brk(0x082CA000)                                 = 0
ioctl(3, ZFS_IOC_OBJSET_STATS, 0x0804464C)      = 0
ioctl(3, ZFS_IOC_DATASET_LIST_NEXT, 0x08045788) Err#3 ESRCH
ioctl(3, ZFS_IOC_SNAPSHOT_LIST_NEXT, 0x08045788) = 0
ioctl(3, ZFS_IOC_OBJSET_STATS, 0x08043484)      = 0
ioctl(3, ZFS_IOC_SNAPSHOT_LIST_NEXT, 0x08045788) Err#3 ESRCH
ioctl(3, ZFS_IOC_DATASET_LIST_NEXT, 0x08046950) = 0
ioctl(3, ZFS_IOC_OBJSET_STATS, 0x0804464C)      = 0
ioctl(3, ZFS_IOC_DATASET_LIST_NEXT, 0x08045788) Err#3 ESRCH
ioctl(3, ZFS_IOC_SNAPSHOT_LIST_NEXT, 0x08045788) = 0
ioctl(3, ZFS_IOC_OBJSET_STATS, 0x08043484)      = 0
ioctl(3, ZFS_IOC_SNAPSHOT_LIST_NEXT, 0x08045788) Err#3 ESRCH
ioctl(3, ZFS_IOC_DATASET_LIST_NEXT, 0x08046950) = 0
ioctl(3, ZFS_IOC_DATASET_LIST_NEXT, 0x08046950) Err#3 ESRCH
ioctl(3, ZFS_IOC_SNAPSHOT_LIST_NEXT, 0x08046950) = 0
ioctl(3, ZFS_IOC_OBJSET_STATS, 0x0804464C)      = 0
ioctl(3, ZFS_IOC_SNAPSHOT_LIST_NEXT, 0x08046950) Err#3 ESRCH
ioctl(3, ZFS_IOC_OBJSET_STATS, 0x08045814)      = 0
ioctl(3, ZFS_IOC_DATASET_LIST_NEXT, 0x08046950) = 0
ioctl(3, ZFS_IOC_DATASET_LIST_NEXT, 0x08046950) = 0
ioctl(3, ZFS_IOC_OBJSET_STATS, 0x0804464C)      = 0
ioctl(3, ZFS_IOC_DATASET_LIST_NEXT, 0x08045788) = 0
ioctl(3, ZFS_IOC_OBJSET_STATS, 0x08043484)      = 0

Let me know if something else is needed to figure out what is going on.

The system is an x4600 running Solaris 10, with a fairly recent patch set.

bash-3.00# uname -a
SunOS gdo-node-1 5.10 Generic_125101-07 i86pc i386 i86pc

bash-3.00# ps -ef | grep zfs
    root 17593 17438   0   Jul 10 pts/3     605:57 zfs receive
gdo-node-2/pcmdi/ftp/pub-backup
    root 25972 18474   0 17:04:03 pts/4       0:00 truss zfs list
    root 25968 18474   0 17:03:12 pts/4       0:08 zfs list
    root 26086 17438   0 17:31:55 pts/3       0:00 grep zfs
    root 25926 23059   0 16:55:44 pts/5       0:12 zfs list
    root 25973 25972   0 17:04:03 pts/4       0:08 zfs list


David
 
 
This message posted from opensolaris.org

Well, the zfs receive process finally died, and now my zfs list works just fine.

If there is a better way to capture what is going on, please let me know and I
can duplicate the hang.

David
 
 
This message posted from opensolaris.org

David Smith wrote:
> Well, the zfs receive process finally died, and now my zfs list works just
> fine.
> 
> If there is a better way to capture what is going on, please let me know
> and I can duplicate the hang.
If you can take a crash dump while it is "hung" and make it available,
we can
take a look.

--matt

Hello David,

Saturday, July 14, 2007, 2:01:23 AM, you wrote:

DS> Well, the zfs receive process finally died, and now my zfs list works
just fine.

DS> If there is a better way to capture what is going on, please let
DS> me know and I can duplicate the hang.


I can observer similar things from time to time, other commands also
hung like ''zpool status'' of ''zfs list'', etc.
sometimes it takes dozen
minutes sometimes even hours after they complete...


-- 
Best regards,
 Robert                            mailto:rmilkowski at task.gda.pl
                                       http://milek.blogspot.com