[SERVER-12738] Add flush time logging on per file basis as well Created: 14/Feb/14  Updated: 06/Dec/22  Resolved: 11/Nov/16

Status: Closed
Project: Core Server
Component/s: Storage
Affects Version/s: None
Fix Version/s: None

Type: Improvement Priority: Trivial - P5
Reporter: Anil Kumar Assignee: Backlog - Storage Execution Team
Resolution: Incomplete Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Assigned Teams:
Storage Execution
Participants:

 Description   

It would be useful to have following per file based flush time:

  • In non-verbose mode, log for flush latency beyond threshold for each file
  • In verbose mode, log for flush latency irrespective of threshold for each file
  • In verbose mode, log for overall flush latency irrespective of threshold.

I believe the overall threshold currently is set to 10000ms, I would ideally assume if individual files are beyond the weighted average of their respective size w.r.t global threshold, it should log it.

Since, this is only every 60 seconds and should have minimal impact, first higher verbosity level above the default logging level should be good for logging such information or if you have any specific suggestion that should be fine.

This level of detail might be useful depending on if files are spread across mount/symlinks or on other devices. If it is taking a long time to flush files it might help to know which dbs/files are having issues.



 Comments   
Comment by Scott Hernandez (Inactive) [ 14/Feb/14 ]

Verbose mode is really a level, starting at 1 and going to 5/10.

Do you have a suggestion for a level or time threshold per file/db?

Not sure what you mean by update/delete randomness and how this is related to this logging enhancement request, perhaps best to remove from description...

Generated at Thu Feb 08 03:29:28 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.