-
Type: Task
-
Resolution: Done
-
Priority: Major - P3
-
None
-
Affects Version/s: 2.7.0
-
Component/s: Error Handling, Read Operations
-
Environment:MongoDB Replicaset v4.0.0, 3 Nodes
Not affected: version 3.6.3
Hi,
we've upgraded our replicaset to mongo db version 4.0.0.
One application has write errors every 10 minutes since the upgrade. After downgrading they disappear again.
We did some tests (with/without authentication) and finally we were able to create a small demo application: https://github.com/alvillain/TestMongo/tree/master/TestMongo2.0 to reproduce the issue.
It is a .Net Core Application just reading data in an endless loop. Every 10 minutes we see one write error. The db/collection does not have any validation rules.
Interesting: If the application is stopped the write error will occur one last time at the end of the 10m intervall as usual.
We were seeing the write errors in our monitoring. From mongo db shell you can get the value with:
(db.runCommand( { serverStatus: 1 } )).metrics.getLastError
{ "wtime" : { "num" : 145, "totalMillis" : 302 }, "wtimeouts" : NumberLong(0) }
In the log file of mongodb we were not able to find some hints about the issue. Probably we need to change the logging level: Which module is interesting in this case?
I think more information could be needed and we are happy to provide them.