[SERVER-24300] Fatal Assertion after unclean shutdown Created: 26/May/16 Updated: 14/Jul/16 Resolved: 03/Jun/16 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Storage |
| Affects Version/s: | 3.0.12 |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Major - P3 |
| Reporter: | Karim Zamani | Assignee: | Kelsey Schubert |
| Resolution: | Done | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||
| Operating System: | ALL | ||||
| Participants: | |||||
| Description |
|
After an unclean shutdown, database aborts with Fatal Assertion 17441 when repair is attempted. Database starts up but aborts with certain queries. Environment:
Assertion backtrace:
|
| Comments |
| Comment by Kelsey Schubert [ 03/Jun/16 ] |
|
Hi kzamani, You may benefit from reviewing our documentation on recovering data following an unclean shutdown. If mongod --repair is unable to to recover your data files, you may use mongodump --repair in your repair attempt which uses a more aggressive data-recovery algorithm. However, this procedure may produce a large amount of duplicated documents that would have to be manually resolved. Please note that SERVER project is for reporting bugs or feature suggestions for the MongoDB server. The best place for this discussion would be on the mongodb-user group. If you have any additional questions, please follow up on your post there. Thank you, |
| Comment by Karim Zamani [ 27/May/16 ] |
|
Currently the only option we have for resync'ing is to take a copy of /data/db while the database is active (which will result in the same issue). For our purposes, we are OK losing some data but we'd like the database to work. Is there way to repair (including discarding bad data) or force the database to ignore corrupt data? This will be greatly appreciated. |
| Comment by Ramon Fernandez Marina [ 27/May/16 ] |
|
kzamani, it seems that the data in this node has become corrupted. The best way forward is to resync from a healthy node. |