[SERVER-8381] Listener timer inaccuracies on non-Linux Created: 29/Jan/13 Updated: 29/Mar/13 Resolved: 29/Mar/13 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | None |
| Affects Version/s: | 2.3.2 |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Major - P3 |
| Reporter: | Eric Milkie | Assignee: | Unassigned |
| Resolution: | Duplicate | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||||||
| Operating System: | Windows | ||||||||||||
| Participants: | |||||||||||||
| Description |
|
On my Windows VirtualBox machine, getElapsedTimeMillis() shows that 640ms pass during a wall clock period of 1000ms. This is rather inaccurate, and this is because we are estimating the select() period on Windows using hardcoded values. This causes problems for code trying to measure long periods of time with this. For example, mongos cursor timeout code is hoping to time out cursors after ten minutes; instead, on Windows, it will time out in 15.6 minutes (this was causing a unit test failure on Windows). |
| Comments |
| Comment by Eliot Horowitz (Inactive) [ 29/Mar/13 ] |
| Comment by Eric Milkie [ 29/Mar/13 ] |
|
Apparently Ubuntu 10.04 LTS is also very inaccurate although it's unclear why. It can't even get past the dbadmin.js test (which is a coarse test of the Listener timer) |
| Comment by auto [ 29/Jan/13 ] |
|
Author: {u'date': u'2013-01-29T18:09:40Z', u'email': u'milkie@10gen.com', u'name': u'Eric Milkie'}Message: |