[SERVER-46446] Log time spent waiting for remote operations Created: 27/Feb/20 Updated: 29/Oct/23 Resolved: 20/Jan/23 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Diagnostics, Querying |
| Affects Version/s: | None |
| Fix Version/s: | 6.3.0-rc0, 6.0.8 |
| Type: | Improvement | Priority: | Major - P3 |
| Reporter: | Josef Ahmad | Assignee: | Yoon Soo Kim |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | qexec-team, query-offsite | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||||||||||||||||||||||
| Assigned Teams: |
Query Execution
|
||||||||||||||||||||||||||||
| Backwards Compatibility: | Fully Compatible | ||||||||||||||||||||||||||||
| Backport Requested: |
v6.0, v5.0, v4.4
|
||||||||||||||||||||||||||||
| Sprint: | Query 2020-03-23 | ||||||||||||||||||||||||||||
| Participants: | |||||||||||||||||||||||||||||
| Case: | (copied to CRM) | ||||||||||||||||||||||||||||
| Description |
|
This ticket is the find() counterpart of |
| Comments |
| Comment by Githook User [ 20/Jun/23 ] |
|
Author: {'name': 'Yoonsoo Kim', 'email': 'yoonsoo.kim@mongodb.com', 'username': 'yun-soo'}Message: (cherry picked from commit be3f6ec23c5cbc4e4b7e563d50551fb7e2c7340b) |
| Comment by Githook User [ 20/Jan/23 ] |
|
Author: {'name': 'Yoonsoo Kim', 'email': 'yoonsoo.kim@mongodb.com', 'username': 'yun-soo'}Message: |
| Comment by Bernard Gorman [ 13/Jan/23 ] |
|
yoonsoo.kim@mongodb.com: yes, I think that's a reasonable approach. I don't think it will be very diagnostically useful for most non-cursor commands, since they typically involve little or no work on mongoS and any excessive execution time therefore implies a problem on the shards. But there may certainly be cases where it's helpful, and at the very least this field would allow engineers to confirm at a glance that the remote op is the source of the issue. Fortunately, we decided to give the field the extremely generic name remoteOpWaitMillis in the original patch, so that we could accommodate this approach later |
| Comment by Yoon Soo Kim [ 12/Jan/23 ] |
|
Hi bernard.gorman@mongodb.com, justin.seyster@mongodb.com, Sorry for pinging you about an old issue. I'm looking at this ticket and related code and the previous CR. I think it would be simpler to log a generic remoteOpTimeWaitMillis for any command which contacts a remote shard using AsyncRequestsSender (Bernard's alternative solution). My rationales are
LMK what you think. If we can agree on a solution, I'll implement it soon. |
| Comment by Bernard Gorman [ 27/Mar/20 ] |
|
After some investigation, and consistent with the discussion on the original CR here, there are some complications in doing this for find which means that we can't get this for free as a result of Moving this to Q2 Quick Wins for further consideration. |