Closed wdroste closed 2 years ago
Thanks for reporting this.
We have tests that cover similar cases and it would be interesting to see if those can be improved.
However, I could not reproduce so far. Here is what I did:
try (
Session session = driver.session( sessionConfig ) )
{
session.writeTransaction( tx ->
{
var result = tx.run( "CREATE (n:Testing) RETURN n" );
result.consume();
return null;
} );
}
Would you be able to do the following please?
Let me see what i can do, because we don't usually take beta-drivers, what's the ETA on release?
There is a GA version already, just use 4.4.1.
Our customer is experiencing a problem with our App and Neo4j 4.2.x Enterprise, using a 3-node Neo4j Causal Cluster, all on RHEL 7.
When the Neo leader node dies and a new leader is elected, the app ui hangs, and the following message appears in log over and over for a while, then the exceptions start:
2021-10-20 13:38:59.196 WARN 865 — [ Neo4jDriverIO-2-9] [-] org.neo4j.driver.LoadBalancer : Failed to obtain a connection towards address 10.x.x.x:7687, will try other addresses if available. Complete failure is reported separately from this entry.
Java Driver: 4.3.2-4
10.x.x.x is the failed node in this case. I have reproduced this problem today with Neo4j 4.2.11 and Neo4j 4.3.6. I am unable to repro with Java Driver 4.1.1 and Neo4j 4.2.11
This leads me to believe that neo4j failover broke somewhere between 4.1.1->4.3.2. In the problem case, ui never recovers. In the problem case, if you restart app - even with the failed neo node still down - the app comes up and works. In the working case, the app ui recovers after a minute or so.