[SERVER-44735] track metrics for system initiated operations (commands, reads) separately from client initiated ones Created: 19/Nov/19 Updated: 25/May/23 |
|
| Status: | Backlog |
| Project: | Core Server |
| Component/s: | Diagnostics |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Improvement | Priority: | Major - P3 |
| Reporter: | Asya Kamsky | Assignee: | Backlog - Query Optimization |
| Resolution: | Unresolved | Votes: | 0 |
| Labels: | qopt-team | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||||||||||
| Assigned Teams: |
Query Optimization
|
||||||||||||||||
| Participants: | |||||||||||||||||
| Description |
|
Currently serverStatus splits some operations to allow tracking them separate if they are internally triggered but not all. For instance deletes from TTL thread are tracked separately from "regular" deletes initiated by client. It would be useful to do the same for reads, getMores and commands initiated by system, like secondaries reading the oplog, automation agent running commands, etc. |
| Comments |
| Comment by Davis Haupt (Inactive) [ 12/Oct/21 ] |
|
I was wondering if there could be more clarity in this ticket on what metrics need to be split out into client-initiated vs system-initiated counts. Should part of this ticket include evaluating which metrics need to be split out? |
| Comment by Bruce Lucas (Inactive) [ 19/Nov/19 ] |
|
We had discussed something like this with regard to the flow control project where we felt it might be useful to exempt "system" operations from flow control on the theory that a) they generally shouldn't need to be throttled, and b) throttling them could be harmful to system health. However no one could come up with a good way of separating system operations from client operations. So it may be difficult to do this; or if a way could be found to do so in this context, that maybe could be applied to flow control. |