Bug 721122
| Summary: | lvremove cannot remove "open" snapshot | ||||||
|---|---|---|---|---|---|---|---|
| Product: | Red Hat Enterprise Linux 6 | Reporter: | Brian Wheeler <bdwheele> | ||||
| Component: | lvm2 | Assignee: | Peter Rajnoha <prajnoha> | ||||
| Status: | CLOSED DUPLICATE | QA Contact: | Corey Marthaler <cmarthal> | ||||
| Severity: | high | Docs Contact: | |||||
| Priority: | unspecified | ||||||
| Version: | 6.1 | CC: | agk, dwysocha, heinzm, jbrassow, mbroz, prajnoha, prockai, thornber, zkabelac | ||||
| Target Milestone: | rc | ||||||
| Target Release: | --- | ||||||
| Hardware: | x86_64 | ||||||
| OS: | Linux | ||||||
| Whiteboard: | |||||||
| Fixed In Version: | Doc Type: | Bug Fix | |||||
| Doc Text: | Story Points: | --- | |||||
| Clone Of: | Environment: | ||||||
| Last Closed: | 2011-08-05 17:21:30 UTC | Type: | --- | ||||
| Regression: | --- | Mount Type: | --- | ||||
| Documentation: | --- | CRM: | |||||
| Verified Versions: | Category: | --- | |||||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||
| Embargoed: | |||||||
| Attachments: |
|
||||||
Do you have udisks package installed? If yes, then this is another instance of the bug #570359. We already have a fix for that, but it only works for intenal device opens within an LVM command. This fix should appear in next lvm2 build that is targeted for RHEL 6.2. But if you open a device for read-write, closing it afterwards and then immediately run lvremove on that, you can get into a race because there's an open from within udev rules based on the watch rule. We have no way to synchronize with such events directly. What you can do to save the situation is to use "udevadm settle" in between such commands (this could have a perfomance impact since with udevadm settle you wait for *all* devices being processed in the system to settle down, not just the one you're processing). Is it the issue with udisks as mentioned in that bug #570359? You can test it quickly as noted in https://bugzilla.redhat.com/show_bug.cgi?id=570359#c10. I'm testing it now. I commented the line per bug 570395 comment 10 and issued a udevadm control --reload-rules. I also added a "udevadm settle" just before the lvremove. It seems to have fixed it, so this is probably a dupe. (In reply to comment #4) > It seems to have fixed it, so this is probably a dupe. Thanks for trying it out. So I'll close this one. *** This bug has been marked as a duplicate of bug 570395 *** *** This bug has been marked as a duplicate of bug 570359 *** |
Created attachment 512728 [details] an online fsck check Description of problem: I've written an online fsck which creates snapshot volumes, but I cannot remove the snapshot volume reliably: it shows up as open even though nothing should be using it. There are monitoring messages in dmesg, but disabling monitoring makes no difference. Version-Release number of selected component (if applicable): lvm2-2.02.83-3.el6.x86_64 How reproducible: It doesn't happen on every lvremove, but often enough that I cannot put the script into cron. Steps to Reproduce: 1. run "online_fsck -a -v" script 2. observe process -- when destroying snapshot volume it may fail 3. even manually trying to remove it will fail, although after time it does allow removal. various combinations of vgscan, lvscan, starting or stopping lvm2-monitor and sync hasn't improved or degraded the situation. Actual results: logical volume 'stuck' in open state. Expected results: removal of snapshot volumes Additional info: I've attached my script that triggers it, but I suspect it could be triggered by just entering the commands by hand. This is the script I'm using on my servers, but I don't make any promises that it won't mess up your machine or run over your pet goldfish. The core commands it is using are: lvcreate -s -l 5%ORIGIN -n $lv-snap $vg/$lv # create lv snapshot (nice e2fsck -p $FSCK_OPTS /dev/$filesystem-snap) && e2fsck -fy -Fttv -C0 /dev/$filesystem-snap # the filesystem check tune2fs -C0 -T now /dev/$filesystem # if it has an ok fsck lvremove -f $vg/$lv-snap # this is what fails