[ https://issues.apache.org/jira/browse/IGNITE-7832?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Andrew Mashenkov updated IGNITE-7832: ------------------------------------- Description: Steps to reproduce: # Start 3 nodes with Partitioned cache with no backups. Register event handlers for EVT_CACHE_REBALANCE_PART_DATA_LOST event. PartitionLossPolicy should be smth other then IGNORED (which is default). # Put some data and make sure rebalance is finished. # Kill first node. I observe partition lost events on 2-nd node only, however, some of lost partitions should be owned by 3-rd node regarding a new topology. Cache.lostPartitions() returns non-empty resultset on 2-nd node only. # Kill 2-nd node. No partition lost events will be fired on 3-rd node. However data is obviously lost. Cache.lostPartitions() returns empty resultset. Looks like partition lost events mechanics is broken and should be fixed. was: Steps to reproduce: # Start 3 nodes with Partitioned cache with no backups. Register event handlers for EVT_CACHE_REBALANCE_PART_DATA_LOST event. PartitionLossPolicy should be smth other then IGNORED (which is default). # Put some data and make sure rebalance is finished. # Kill first node. I observe partition lost events on 2-nd node only however, some of them should be owned by 3-rd node according new topology. Cache.lostPartitions() returns non-empty resultset on 2-nd node only. # Kill 2-nd node. No partition lost events will be fired on 3-rd node. However data is obviously lost. Cache.lostPartitions() returns empty resultset. Looks like partition lost events mechanics is broken and should be fixed. > Partition lost events looks broken. > ----------------------------------- > > Key: IGNITE-7832 > URL: https://issues.apache.org/jira/browse/IGNITE-7832 > Project: Ignite > Issue Type: Task > Components: cache > Reporter: Andrew Mashenkov > Priority: Major > Fix For: 2.5 > > > Steps to reproduce: > # Start 3 nodes with Partitioned cache with no backups. > Register event handlers for EVT_CACHE_REBALANCE_PART_DATA_LOST event. > PartitionLossPolicy should be smth other then IGNORED (which is default). > # Put some data and make sure rebalance is finished. > # Kill first node. > I observe partition lost events on 2-nd node only, however, some of lost > partitions should be owned by 3-rd node regarding a new topology. > Cache.lostPartitions() returns non-empty resultset on 2-nd node only. > # Kill 2-nd node. > No partition lost events will be fired on 3-rd node. However data is > obviously lost. > Cache.lostPartitions() returns empty resultset. > Looks like partition lost events mechanics is broken and should be fixed. -- This message was sent by Atlassian JIRA (v7.6.3#76005)