[SERVER-18307] Increased lock time Created: 04/May/15  Updated: 05/May/15  Resolved: 05/May/15

Status: Closed
Project: Core Server
Component/s: Concurrency
Affects Version/s: 2.6.9
Fix Version/s: None

Type: Question Priority: Minor - P4
Reporter: Aristarkh Zagorodnikov Assignee: Sam Kleinman (Inactive)
Resolution: Done Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Attachments: PNG File chart.png    
Participants:

 Description   

After upgrade from 2.6.8 to 2.6.9 our monitoring system detected increased lock time for almost all databases. Attaching graph for one of them (upgrade was done on 06.04). While this did not create any real problems yet, all of our databases exhibited this lock time growth and I wonder if it's a real change or maybe it's just an adjustment in lock time calculations/reporting?



 Comments   
Comment by Aristarkh Zagorodnikov [ 05/May/15 ]

Thank you, for your reply. It appears that [election] priorities were updated as a part of the upgrade process, which resulted in primary being moved to another server (the one in question). Please excuse me for the false alert and thank you again.

Comment by Sam Kleinman (Inactive) [ 04/May/15 ]

Hi,

Looking over the changes in 2.6.9, nothing would obviously explain the effect that you're witnessing; however, SERVER-14071 could potentially explain an effect like this. aThe complete list of changes in 2.6.9 is here.

If SERVER-14071 doesn't seem to be at play and if nothing jumps out at you from the changelog, the following questions will help us begin to investigate this issue:

  • Did the other members of the replica set change their syncing targets concurrently with the upgrade?
  • Were there other configuration or application changes that might have corresponded with the upgrade or would have been related to restarting the set member?
  • Did any of the other members of the replica set experience the same change following the upgade or was it limited to this instance?
  • Have you investigated the logs? If you have logs where a system is experiencing higher-than-usual lock time, potentially with log-level 1 enabled, that might help us spot anomalies. You can attach those logs to this ticket.

Regards,
sam

Comment by Aristarkh Zagorodnikov [ 04/May/15 ]

P.S. The second lock time growth is a result of the node becoming primary, it was secondary at the time of upgrade.

Generated at Thu Feb 08 03:47:15 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.