Thanks Val, I just thought the "failoverSafe" flag on semaphore creation would take care of this.
Tested it again with REPLICATED mode, and it works like a charm. With default mode and 2 backups configured, i get a "ClusterTopologyCheckedException: Failed to acquire lock for keys (primary node left grid, retry transaction if possible)" on both remaining nodes, but one of them manages to acquire the semaphore. So I guess that's ok. Thanks again to all of you guys helping me out. Mario -- View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Semaphore-blocking-on-tryAcquire-while-holding-a-cache-lock-tp3031p3097.html Sent from the Apache Ignite Users mailing list archive at Nabble.com.