[SERVER-19989] Segfault in evict_lru.c Created: 17/Aug/15 Updated: 09/Sep/15 Resolved: 28/Aug/15 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | WiredTiger |
| Affects Version/s: | None |
| Fix Version/s: | 3.0.6 |
| Type: | Bug | Priority: | Critical - P2 |
| Reporter: | Ramon Fernandez Marina | Assignee: | Michael Cahill (Inactive) |
| Resolution: | Done | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||||||
| Backwards Compatibility: | Fully Compatible | ||||||||||||
| Operating System: | ALL | ||||||||||||
| Participants: | |||||||||||||
| Description |
|
| Comments |
| Comment by Michael Cahill (Inactive) [ 19/Aug/15 ] |
|
Update: this could have been caused by not backporting https://github.com/wiredtiger/wiredtiger/commit/393344d5d4b436fe3519cb8ab541bab22663553d to 3.0. I have corrected that now by backporting it to the WT mongodb-3.0 branch ready for a new drop into MongoDB 3.0. I will try to trigger that condition in RC1 to verify that this change fixes it. |
| Comment by Michael Cahill (Inactive) [ 19/Aug/15 ] |
|
This issue has been hit again in testing 3.0.6-rc1. That version doesn't include the write barrier change here, but we have no reason to believe that fixes it. We need a repro. |
| Comment by Githook User [ 19/Aug/15 ] |
|
Author: {u'username': u'michaelcahill', u'name': u'Michael Cahill', u'email': u'michael.cahill@mongodb.com'}Message: Merge pull request #2124 from wiredtiger/dhandle-barrier
|
| Comment by Githook User [ 19/Aug/15 ] |
|
Author: {u'username': u'michaelcahill', u'name': u'Michael Cahill', u'email': u'michael.cahill@mongodb.com'}Message: Merge pull request #2126 from wiredtiger/dhandle-barrier-all
|
| Comment by Githook User [ 19/Aug/15 ] |
|
Author: {u'username': u'michaelcahill', u'name': u'Michael Cahill', u'email': u'michael.cahill@mongodb.com'}Message: |
| Comment by Keith Bostic (Inactive) [ 18/Aug/15 ] |
|
michael.cahill, I've looked and I don't see anything else in/around this path. |
| Comment by Michael Cahill (Inactive) [ 18/Aug/15 ] |
|
I've been running workloads to keep eviction busy while constantly creating and discarding handles for hours today without triggering this. Given the timing, my assumption was that it was introduced by The only other thought I've had is that we might somehow be seeing a new handle before the name field has been set (which would make more sense in the sweep server than here), so I've opened https://github.com/wiredtiger/wiredtiger/pull/2124 to add memory barriers before handles are put in shared lists. keith.bostic, any other ideas? |