Description of problem: I see this error sometimes, usually a re-run on cleanup fixes it. Sometimes it is because I was trying to run cleanup while on the nfs mount. This particular time, I ran into a issue where dc-rhev-image (push) command was hung. If the user is running cleanup, then we probably could/should kill any push processes
Forgot to post the cleanup error: err: /Stage[main]/Aeolus::Rhevm::Disabled/Mount[/mnt/rhevm-nfs]/ensure: change from mounted to unmounted failed: Execution of '/bin/umount /mnt/rhevm-nfs' returned 16: umount.nfs: /mnt/rhevm-nfs: device is busy umount.nfs: /mnt/rhevm-nfs: device is busy
making sure all the bugs are at the right version for future queries
} 2011-11-16 18:13:13,432 DEBUG imgfac.builders.BaseBuilder.RHEL6_rhevm_Builder pid(3777) Message: Executing external RHEV-M push command (['/usr/bin/dc-rhev-image', '/tmp/tmpQzGWlu']) 2011-11-16 18:13:13,447 DEBUG paste.httpserver.ThreadPool pid(3777) Message: Added task (0 tasks queued) ^C [root@qeblade27 ~]# aeolus-cleanup notice: /Stage[main]/Aeolus::Conductor::Disabled/Service[conductor-dbomatic]/ensure: ensure changed 'running' to 'stopped' err: /Stage[main]/Aeolus::Profiles::Rhevm::Disabled/Mount[/mnt/rhevm-nfs]/ensure: change from mounted to unmounted failed: Execution of '/bin/umount /mnt/rhevm-nfs' returned 16: umount.nfs4: /mnt/rhevm-nfs: device is busy umount.nfs4: /mnt/rhevm-nfs: device is busy Still seeing this with yesterdays rpms [root@qeblade27 ~]# rpm -qa | grep aeolus aeolus-conductor-0.6.0-0.20111114172747gitde62906.fc15.noarch rubygem-aeolus-cli-0.1.0-3.20111114114625git42b14c0.fc15.noarch aeolus-conductor-daemons-0.6.0-0.20111114172747gitde62906.fc15.noarch rubygem-aeolus-image-0.1.0-4.20111024205454git6b2b696.fc15.noarch aeolus-conductor-doc-0.6.0-0.20111114172747gitde62906.fc15.noarch aeolus-configure-2.3.0-0.20111111160331git9bde372.fc15.noarch aeolus-all-0.6.0-0.20111114172747gitde62906.fc15.noarch
Dave, I'm having trouble figuring out a way to kill dc-rhev-image. A kill or kill -9 has no affect on it. I looked into nfs mount options and there isn't one to make nfs operations interruptible. There was intr but that has been deprecated since kernel 2.6.25. Since the bug was opened, rhev pushes do have a specified timeout, currently at 30 mins. So that may have helped with the hung process you experienced before. I'm going to add some iwhd folks to the cc list to see if they have any suggestions. - Richard
Hi Richard, From what I read about the deprecation of the "intr" mount option, we are supposed to be able to use kill -9 to interrupt a pending NFS operation. Since that has no effect, you may want to talk to NFS or kernel people.
steved said that interrupting nfs operations is not recommended because it can lead to data corruption. I suggest we leave the it unmounted. I can add a check to the unmount operation to check if there are any dc-rhev-image processes active? If there are, don't unmount. Dave, how does that sound to you?
patch pushed to configure commit 83993ba03907354a552a060de3f664fd04fa1167
Available in aeolus-configure-2.5.0-0.20111222093431gitd3ed726
This one was blocked by bug 788644 but that dependency is no longer valid. cleanup is now down using 'aeolus-cleanup -p rhevm' good 2 go in aeolus-configure-2.5.0-15.el6.noarch
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. http://rhn.redhat.com/errata/RHEA-2012-0586.html