[SERVER-43311] Uncommon bug makes logical session cache not refresh Created: 13/Sep/19  Updated: 06/Dec/19  Resolved: 06/Dec/19

Status: Closed
Project: Core Server
Component/s: None
Affects Version/s: 4.0.12
Fix Version/s: None

Type: Bug Priority: Major - P3
Reporter: Iacobescu Tudor Assignee: Danny Hatcher (Inactive)
Resolution: Incomplete Votes: 2
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Operating System: ALL
Steps To Reproduce:

Unknown, but we suspect it is caused by or related to high loads and/or database crashes.

Participants:

 Description   

A problem that has occurred on a replica set twice thus far - once in production, once during testing. For some reason, sometimes the logical session cache stops refreshing against the main session store, collecting a backlog of logical sessions until it hits the 1 million session cap, which causes queries to start to fail with a 261: TooManyLogicalSessions error.

The tests we did to try and pinpoint the cause of the bug (and with which we caused it again) were using the NodeJS 3.3.2 driver (for Node 10.16.3), and the Python 3.9.0 driver (for Python 3.7.4)  respectively.

The test environment was a 4.0.12 replicaset with one primary, one secondary and an arbiter. 

 



 Comments   
Comment by Danny Hatcher (Inactive) [ 08/Nov/19 ]

The ticket you linked is behavior that is working as designed. In order to confirm whether or not that was the situation, we'd need the logs and "diagnostic.data" that I mentioned in my previous comment. Have you experienced the issue again since the original occurrence?

Comment by Iacobescu Tudor [ 17/Sep/19 ]

https://jira.mongodb.org/browse/SERVER-43312 
May be related to this.

Comment by Iacobescu Tudor [ 17/Sep/19 ]

Unfortunately, we have not managed to reproduce the bug again yet. We will continue trying, and will supply the logs when it happens again.

Comment by Danny Hatcher (Inactive) [ 16/Sep/19 ]

In order for us to investigate, please provide the following to our Secure Uploader. Only MongoDB engineers will be able to access the files.

  • mongod log file from Primary covering from a restart until you first receive the error
  • "diagnostic.data" folder from Primary for the same timeframe
Generated at Thu Feb 08 05:02:50 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.