[SERVER-21672] Mongod consumes 100% cpu when drive out of space Created: 25/Nov/15 Updated: 07/Dec/15 Resolved: 07/Dec/15 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | None |
| Affects Version/s: | 3.2.0-rc3 |
| Fix Version/s: | None |
| Type: | Improvement | Priority: | Minor - P4 |
| Reporter: | Nick Judson | Assignee: | Kelsey Schubert |
| Resolution: | Done | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Attachments: |
|
| Participants: |
| Description |
|
When mongod runs out of drive space on windows, it may sit and spin consuming 100% cpu. Although I haven't confirmed if this happens when the out of space scenario occurs, it seems to happen when a shutdown is initiated with no available space. |
| Comments |
| Comment by Kelsey Schubert [ 07/Dec/15 ] |
|
Thank you for testing this behavior on 3.2.0-rc4 as well as 3.2.0-rc3. I am closing this ticket as gone away since you reported a different result in the more recent version. SERVER-13811 is scheduled to improve how mongoD handles running out of disk space. Please feel free to watch or vote for it. Kind regards, |
| Comment by Nick Judson [ 26/Nov/15 ] |
|
I got a different result after upgrading to RC4 (see the screen cap). |
| Comment by Nick Judson [ 25/Nov/15 ] |
|
I'm not sure - I only noticed it after ctrl-c. I can re-run the test tonight and let you know. I would expect it to keep trying (sleeping between tries) until a shutdown is triggered, at which point, there is a max amount of time which will elapse before self-destruct (with an appropriate error). I also wouldn't expect the spinning cpu, just a try/sleep cycle. |
| Comment by Mark Benvenuto [ 25/Nov/15 ] |
|
Was the mongod process spinning before shutdown started or only after shutdown was started by Ctrl-C? How do you expect mongod to behave in this case? If we are out of disk space, we may not be able to flush all pending changes to disk for instance. |