Andrey, thanks for getting back. I am attaching the stack trace. Don't think the cause is a deadloc, but the trace is long so maybe I am missing out something, let me know if you find something useful.
We cannot ourselves reproduce this issue as there are no errors on the prior successful update. It feels like the txn was marked successful, but on one of the keys the lock was not released. and later when we try to access the key, its locked, hence the exceptions. No messages in the logs for long running txns or futures. By killing the node that holds the key, the lock is not released. Is there a way to query ignite to see if the locks are being held on a particular key? Any code we can run to salvage such locks? Any other suggestions? Thanks, Binti -- View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Pessimistic-TXN-did-not-release-lock-on-a-key-all-subsequent-txns-failed-tp10536p10764.html Sent from the Apache Ignite Users mailing list archive at Nabble.com.
