Join GitHub today
GitHub is home to over 36 million developers working together to host and review code, manage projects, and build software together.Sign up
Seeing multiple concurrent locks on a distributed lock #168
I'm seeing a case where the same lock can be locked on two different machines at the same time. I perform a ping before attempting the distributed lock to avoid attempting to lock in a split brain scenario.
Hazelcast version: Hazelcast Community Edition 2.0.3 (20120418)
I looked at the release notes for version Hazelcast v2.1, and don't see any fixes that look relevant.
In the logs below, each line is tagged with the machine name, "dg0" or "dg1".
I had the Hazelcast logging set to FINEST, but I don't really understand the Hazelcast logging messages or how they might relate to my problem. Any help here would be appreciated.
Group discussion for this issue:
I commented on issue #267 that I am still seeing that issue in 2.4.1. I have updated to 2.5, and am still seeing both issues. I was able to hack around #267 pretty easily (manually tracking which thread owns the lock, and catching the exception when it is not valid). This issue still happens and exists for me, although I am also able to detect this and work around out in a bit of a hacky way. It seems that when one node fails in the cluster, this issue may be more likely to occur.
Just wanted to make sure it was documented that both this, and #267 appear to still be defects.