Hello experts, Cluster details: Running Ignite 2.4.0 with 3 servers and 4 clients.
Observed transaction timeouts after a server node restarted. Upon further investigation, it was due to PME failures and also saw that there were some long running transactions on a cache for the same key for more than 9 days (duration=840832760ms). 2 client nodes had created the write (delete the key-value) transaction at around the same time with timeout set to 50ms. One of the transaction was stuck in state=PREPARED and the other in state=COMMITTING. Is there any known issue in 2.4.0 which causes this deadlock? Why did Ignite not honor the timeout value of 50ms and bail out? Regards, rc -- Sent from: http://apache-ignite-users.70518.x6.nabble.com/
