[SERVER-20010] Segfault while dropping an index that failed to build Created: 18/Aug/15 Updated: 06/Dec/22 Resolved: 08/Dec/17 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Index Maintenance |
| Affects Version/s: | 2.6.9 |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Major - P3 |
| Reporter: | Kyle Erf | Assignee: | Backlog - Query Team (Inactive) |
| Resolution: | Done | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Environment: |
3 node replica set |
||
| Issue Links: |
|
||||||||
| Assigned Teams: |
Query
|
||||||||
| Operating System: | Linux | ||||||||
| Participants: | |||||||||
| Description |
|
We have a user hitting a situation where they 1. started a background build on index A, a sparse index
3. this failure closed the collection's cursors, which in turn also caused the build of index A to fail 4. the background index A build fails, but the index is still listed in db.col.getIndexes() 5. the user attempts to drop to index A with the intention of rebuilding it, but on removing the index, they get
|
| Comments |
| Comment by Ramon Fernandez Marina [ 11/Sep/15 ] | |||||||||||||||||||||||||||||||||||||||||||||||
|
Sorry you're running into this issue klimashkin. This ticket is scheduled for the next 2.6 dot release, but there's currently no target date for that. As Eric mentions above this bug is not present on the MongoDB 3.0 series, so if this issue is critical for you you may want to consider upgrading to MongoDB 3.0.6. Regards, | |||||||||||||||||||||||||||||||||||||||||||||||
| Comment by Paul Klimashkin [ 11/Sep/15 ] | |||||||||||||||||||||||||||||||||||||||||||||||
|
Really critical issue. Our mongodb crashed twice a day after update to 2.6.11 | |||||||||||||||||||||||||||||||||||||||||||||||
| Comment by Eric Milkie [ 18/Aug/15 ] | |||||||||||||||||||||||||||||||||||||||||||||||
|
The work for | |||||||||||||||||||||||||||||||||||||||||||||||
| Comment by J Rassi [ 18/Aug/15 ] | |||||||||||||||||||||||||||||||||||||||||||||||
|
Looks like client->curop() is returning null in IndexCatalog::killMatchingIndexBuilds(), which sounds like a vaguely familiar race condition. Related to addr2line output:
|