sessionExpired errors when using the driver from nodeJS to aura
jroith opened this issue · 7 comments
I'm still getting random sessionExpires errors when connecting to aura even though I have set the max connection timeout to 9 minutes. Wouldn't it be better if the driver would transparently retry instead of failing? And in any case, why can it not just retry with a different or even a new connection?
Also it would be nice to maintain the connections in the pool more actively to prevent this problem.
Hi @jroith,
we have methods in the driver which internally retries when some cluster failures happens. The methods are:
- Driver.executeQuery(Manual). This one should be used to run a single query when you don't need to have control over the cursor/stream.
- Session.executeRead(Manual). For read queries.
- Session.executeWrite(Manual). For write queries.
Are you using these methods? Could you share some code samples?
This happens from the neo4j GraphQL library and they are using these methods. I will add more logging and try to report more information. Unfortunately the issue is difficult to reproduce and only occurs sometimes.
Yes, it's getting worse and feels unpredictable for long-running transactions. I'm currently considering running the community or enterprise version instead of aura.
Can you share some logs?
Not recently. I have just upgraded to 5.17 and will keep monitoring the issue.