Hi I have a script that creates accounts and zfs file systems for each
account.  The list of accounts is fairly large ~5200.  I have in the past
been able to run this script and create filesystems for all 5200 accounts.
With snv_107 I get periodic failures while running the script.  primarily
the failure seem to revolve around actions taken after the zfs file systems
are created e.g this is an excerpt from a relevant portion of the script:

${ZFS} create -o mountpoint=${Z_WEBPATH}  ${W_ZFS}
> RTN=$?
> if [ $RTN -ne 0 ]; then
>     Msg="ZFS create ${Z_WEBPATH} in ${U_WEB} failed with return=$RTN
> [$REC]."
>     $ACCTLOG $LOG_ERR "$Msg"
>     . /usr/local/idm/CaptureEnv
>     return $FAIL
> fi
>
> $MKDIR ${Z_WEBPATH}/public_html
> $MKDIR ${Z_WEBPATH}/htaccess
> cp /etc/skel/* ${Z_HOMEPATH}
> cp /etc/skel/.profile ${Z_HOMEPATH}
>
> $CHOWN -R $UNAME:$GID ${Z_HOMEPATH}
> $CHOWN -R $UNAME:$GID ${Z_WEBPATH}
>
>
an example of the errors I see:

ddxxxxx,61622,500,student,David D Dxxxxx
> mkdir: cannot create directory `/srv/web/u0/d/d/ddxxxxx/public_html': No
> such file or directory
> mkdir: cannot create directory `/srv/web/u0/d/d/ddxxxxxhtaccess': No such
> file or directory
> chown: cannot access `/srv/web/u0/d/d/ddxxxxx': No such file or directory


and yet the filesystem exists and is mounted at some later point:

$ zfs list | grep ddxxxxx
> home/ddxxxxx              23.5K  3.00G  23.5K  /srv/home/u0/d/d/ddxxxxx
> web/ddxxxxx                 18K   300M    18K  /srv/web/u0/d/d/ddxxxxx



I can only assume that there is some race condition, that zfs returns before
it has actually mounted the filesystem.  Similar errors also happen
periodically on the other pool.   The script runs correctly initially and
then errors start appearing with increasing frequency, but not
consistently.  The underlying devices of the pool are iscsi volumes so they
may not be the fastest pools, however, I have successfully run this script
without error on previous versions of OS and Solaris 10.

Any ideas?  is there a known issue I should be aware of?

Thanks,  Alastair Neil
_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss

Reply via email to