Has anyone got this error?

[killJoy-03]

System.Reflection.TargetInvocationException: 'Exception has been thrown by the target of an invocation.'
Inner Exception:
InvalidOperationException: Operations that change non-concurrent collections must have exclusive access. A concurrent update was performed on this collection and corrupted its state. The collection's state is no longer correct.

This occurs randomly.
Driver v5.27.0
.NET 9

[RichardIrons-neo4j]

Hi there. I'd like to get to the bottom of this - next time it happens, please could you copy as much of the error as possible and post it here?

[killJoy-03]

Sure

[killJoy-03]

@RichardIrons-neo4j
I have found the pattern

1. We are calling multiple queries (in separate sessions)
   suppose
   query1 (run 50times)
   query2 (run 20times)
   query3 (run 10times)

Now query1 runs successfully
but query2 throws the above error.

All are using default mapping ToListAsync()

Now I gave a custom mapping to query2
Then query3 starts throwing the above error.

Then I gave custom mapping to query3
now it starts throwing the above error in the next query.

[killJoy-03]

All sessions are in read mode.

[killJoy-03]

Is there any chance where driver is updating mappers dictionary internally and failing while concurrent update?

[AndyHeap-NeoTech]

Hi KillJoy. Richard is on vacation for this week, I'll make sure he follows up on his return. It certainly seems like there may be a concurrency issue in the mapping.

A couple of follow on questions to get some more details.

1. Are you creating a new session for each run of each query? Or keeping a session around for all runs of a query (e.g. Query 1 uses the same session for all 50 calls)?
2. Are the queries all mapping to the same object types?

[killJoy-03]

1. all queries are in different sessions.
2. query1(50 times) is related to 1 object. query2(20 times) is related to another object
   In short, each set of queries is related to 1 object.

[killJoy-03]

@AndyHeap-NeoTech @RichardIrons-neo4j any update on this?

[AndyHeap-NeoTech]

Morning @killJoy-03, as it so happens we merged a PR to fix this yesterday evening. We are currently in the process of preparing a 5.28 (LTS) driver release which has a few other bits to go in but should be released in the short term.

[killJoy-03]

@AndyHeap-NeoTech Morning and thanks for the fix, is there any estimate date for this version.
We have a deployment in upcoming days,if this update will take time I have to write custom mapper everywhere.

[RichardIrons-neo4j]

@killJoy-03

This bug has now been fixed, and will be gone in the next release.

PR that fixes the issue: #841