Bug 623790
| Summary: | cmirrord: Failed to open checkpoint: SA_AIS_ERR_NOT_EXIST | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | Red Hat Enterprise Linux 6 | Reporter: | Corey Marthaler <cmarthal> | ||||||||||||||
| Component: | corosync | Assignee: | Steven Dake <sdake> | ||||||||||||||
| Status: | CLOSED CURRENTRELEASE | QA Contact: | Cluster QE <mspqa-list> | ||||||||||||||
| Severity: | urgent | Docs Contact: | |||||||||||||||
| Priority: | urgent | ||||||||||||||||
| Version: | 6.0 | CC: | agk, antillon.maurizio, cluster-maint, coughlan, dwysocha, fdinitto, heinzm, jbrassow, joe.thornber, mbroz, nstraz, prockai, sdake | ||||||||||||||
| Target Milestone: | rc | Keywords: | Regression, TestBlocker | ||||||||||||||
| Target Release: | --- | ||||||||||||||||
| Hardware: | All | ||||||||||||||||
| OS: | Linux | ||||||||||||||||
| Whiteboard: | |||||||||||||||||
| Fixed In Version: | corosync-1.2.3-20.el6 | Doc Type: | Bug Fix | ||||||||||||||
| Doc Text: | Story Points: | --- | |||||||||||||||
| Clone Of: | Environment: | ||||||||||||||||
| Last Closed: | 2010-11-10 22:07:28 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: |
|
||||||||||||||||
|
Description
Corey Marthaler
2010-08-12 19:05:46 UTC
Created attachment 438509 [details]
log from hayes-01
I hit this on another cluster when attempting to start the clvmd init scripts with an existing mirror. Aug 12 10:20:59 grant-03 lvm[2203]: Monitoring mirror device centipede-centi_base for events. Aug 12 10:21:03 grant-03 cmirrord[2141]: [TGcIbDFm] Failed to open checkpoint for 2: SA_AIS_ERR_NOT_EXIST Aug 12 10:21:03 grant-03 cmirrord[2141]: [TGcIbDFm] Failed to export checkpoint for 2 Aug 12 10:21:18 grant-03 kernel: device-mapper: dm-log-userspace: [TGcIbDFm] Request timed out: [9/221] - retrying Aug 12 10:21:33 grant-03 kernel: device-mapper: dm-log-userspace: [TGcIbDFm] Request timed out: [9/224] - retrying Aug 12 10:21:48 grant-03 kernel: device-mapper: dm-log-userspace: [TGcIbDFm] Request timed out: [9/225] - retrying Aug 12 10:22:03 grant-03 kernel: device-mapper: dm-log-userspace: [TGcIbDFm] Request timed out: [9/226] - retrying Aug 12 10:22:06 grant-03 cmirrord[2141]: Sync checkpoint section creation failed: SA_AIS_ERR_NOT_EXIST Aug 12 10:22:06 grant-03 cmirrord[2141]: [TGcIbDFm] Failed to export checkpoint for 1 Aug 12 10:22:18 grant-03 kernel: device-mapper: dm-log-userspace: [TGcIbDFm] Request timed out: [9/227] - retrying Aug 12 10:22:33 grant-03 kernel: device-mapper: dm-log-userspace: [TGcIbDFm] Request timed out: [9/228] - retrying Aug 12 10:22:48 grant-03 kernel: device-mapper: dm-log-userspace: [TGcIbDFm] Request timed out: [9/229] - retrying Aug 12 10:23:03 grant-03 kernel: device-mapper: dm-log-userspace: [TGcIbDFm] Request timed out: [9/230] - retrying Aug 12 10:23:18 grant-03 kernel: device-mapper: dm-log-userspace: [TGcIbDFm] Request timed out: [9/231] - retrying Aug 12 10:23:33 grant-03 kernel: device-mapper: dm-log-userspace: [TGcIbDFm] Request timed out: [9/232] - retrying This seems like an issue that would be related to when cmirror/clvmd scripts are started WRT each other, and how they are re-cycled. Is there a simple reproducer? How often does it happen? The consequences of the bug seem pretty severe, but I'd like more info to gauge the urgency of the issue. Hmmm, just hit this again on test case #9, must be something special with this conversion/creation. --------------------------------------------- TEST CASE=9 base legs 2; convert to 2 legs; base log disk; convert to disk log; active 0; sync=1; pvs=sufficient --------------------------------------------- creating a base mirror on hayes-02 Error locking on node hayes-02: Command timed out Aborting. Failed to activate new LV to wipe the start of it. Error locking on node hayes-03: Command timed out Error locking on node hayes-02: Command timed out Unable to deactivate failed new LV. Manual intervention required. This appears to be a fairly major regression from the previous build. I went back to the 2.02.72-6 build and verified that this test case worked there. This bug need the 6.0 and blocker flag set. I said 'similar to the RHEL5 bug 566799'. :) I'm not positive that's it the same, in fact I'd bet that it's not. This issues didn't appear to exist until the very last build, at which point tests that had previous been working started failing quite easily. Created attachment 439026 [details]
corosync-blackbox from grant-01.lab
Created attachment 439027 [details]
corosync-blackbox from grant-02.lab
Created attachment 439028 [details]
coro
grant-01 was the machine that incurred the following errors: Aug 16 17:04:44 grant-01 cmirrord[13915]: Clean checkpoint section creation failed: SA_AIS_ERR_LIBRARY Aug 16 17:04:45 grant-01 cmirrord[13915]: [vXeHaaML] Failed to export checkpoint for 2 Aug 16 17:04:45 grant-01 cmirrord[13915]: Sync checkpoint section creation failed: SA_AIS_ERR_NOT_EXIST Aug 16 17:04:45 grant-01 cmirrord[13915]: [vXeHaaML] Failed to export checkpoint for 3 Created attachment 439039 [details]
test case which may represent problem
After speaking with Jon, it appears he is getting a CPG_ERR_LIBRARY return code first and then bad things happen.
The test case in attachment #14 [details] is meant to be run as follows:
run cluster on 2 nodes in foreground (aisexec -f)
start ckptbench.c (compiled into openais tree)
it will sleep for 5 seconds. while it is sleeping, press ctrl-z (to background the corosync process). (ctrl-c will gracefully exit)
This triggers a 10 second configuration change where ipc is blocked.
Then the test case tries to do a sectioncreate which returns ERR_LIBRARY.
The root of the problem is that openais follows a design pattern where ipc messages are not responded to during configuration changes to enable correctness.
Our ipc system uses a semaphore timeout operation. This is to allow ipc clients to detect that the server has failed in some way and return an error.
Created attachment 439045 [details]
patch to fix problem
missing %patch from spec file caused qa failure on -19. This bug appears to be fixed in the latest build, however I'm going to let the tests run all night be for marking verified. Tests ran all night (albeit rather slowly). Marking verified in the latest rpms. 2.6.32-59.1.el6.x86_64 corosync-1.2.3-20.el6.x86_64 lvm2-2.02.72-8.el6 BUILT: Wed Aug 18 10:41:52 CDT 2010 lvm2-libs-2.02.72-8.el6 BUILT: Wed Aug 18 10:41:52 CDT 2010 lvm2-cluster-2.02.72-8.el6 BUILT: Wed Aug 18 10:41:52 CDT 2010 udev-147-2.22.el6 BUILT: Fri Jul 23 07:21:33 CDT 2010 device-mapper-1.02.53-8.el6 BUILT: Wed Aug 18 10:41:52 CDT 2010 device-mapper-libs-1.02.53-8.el6 BUILT: Wed Aug 18 10:41:52 CDT 2010 device-mapper-event-1.02.53-8.el6 BUILT: Wed Aug 18 10:41:52 CDT 2010 device-mapper-event-libs-1.02.53-8.el6 BUILT: Wed Aug 18 10:41:52 CDT 2010 cmirror-2.02.72-8.el6 BUILT: Wed Aug 18 10:41:52 CDT 2010 *** Bug 627297 has been marked as a duplicate of this bug. *** Red Hat Enterprise Linux 6.0 is now available and should resolve the problem described in this bug report. This report is therefore being closed with a resolution of CURRENTRELEASE. You may reopen this bug report if the solution does not work for you. *** Bug 627297 has been marked as a duplicate of this bug. *** |