[SERVER-7888] E11000 duplicate key error index while stopping the Balancer Created: 10/Dec/12 Updated: 08/Mar/13 Resolved: 24/Feb/13 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Sharding |
| Affects Version/s: | 2.0.8 |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Major - P3 |
| Reporter: | Michael Tewner | Assignee: | David Hows |
| Resolution: | Incomplete | Votes: | 0 |
| Labels: | DuplicateKey, balancer | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Environment: |
Linux 64-bit |
||
| Attachments: |
|
||||||||
| Issue Links: |
|
||||||||
| Operating System: | ALL | ||||||||
| Participants: | |||||||||
| Description |
|
Our MongoDB Backup script (python) disables the balancer using: self.config_db.set_lasterror_options(fsync=True) , { '$set' : { 'stopped' : True}}, True, False, True) About 30% of the time, the following exception is thrown: 06:41:18 Traceback (most recent call last): , { '$set' : { 'stopped' : bol } }, True, False, True) After a few attempts, the script succeeds in disabling the balancer. |
| Comments |
| Comment by David Hows [ 18/Dec/12 ] |
|
Hi Michael, Given what you have said I would suspect Especially given that the error is from an unrelated collection Database.m_sequences. Can you upload the mongos logs and if possible the logs of both primaries? I would like to try and confirm that this is the case. Cheers, David |
| Comment by Michael Tewner [ 17/Dec/12 ] |
|
I should add, though, that I was mistaken - The environment is made up of 2 shards of replica sets, but each replica set only has a Primary node. Could that be relevant? |
| Comment by Michael Tewner [ 16/Dec/12 ] |
|
Is it possible that I'm getting { 'err': None } ? 11:34:20 {u'ok': 1.0, u'err': None, u'singleShard': u'devshare01srv02:20001,devshare02srv02:20001,devshare04:20001', u'connectionId': 11942, u'n': 1, u'updatedExisting': True, u'waited': 13}11:34:20 Traceback (most recent call last): , { '$set' : { 'stopped' : bol } }, True, False, True) |
| Comment by David Hows [ 14/Dec/12 ] |
|
Hi Michael, Would you be able to output the whole structure of the getLastError object and attach to the ticket? Cheers, David |