Ivan Artukhov created IGNITE-9144: ------------------------------------- Summary: A client node leaving a grid may trigger the wrong message about coordinator change in the logs Key: IGNITE-9144 URL: https://issues.apache.org/jira/browse/IGNITE-9144 Project: Ignite Issue Type: Bug Reporter: Ivan Artukhov
The issue was introduced by https://issues.apache.org/jira/browse/IGNITE-8738. Suppose we have a grid with X server nodes and Y client nodes. Server nodes are restarted periodically while client nodes are left untouched. In this case *order* of current coordinator might be greater than *order* of any client node. Then when some client node leaves the grid, we will erroneously print the *Coordinator changed* message with *client* node being the previous coordinator. E.g.: {noformat} [2018-07-19 14:55:28,897][INFO ][disco-event-worker-#61] Node left topology: TcpDiscoveryNode [id=7240957f-a51b-452d-bfc8-420e8ef9ea68, addrs=[127.0.0.1, 172.17.0.1, 172.25.1.15], sockAddrs=[/172.17.0.1:0, /127.0.0.1:0, lab15.gridgain.local/172.25.1.15:0], discPort=0, order=16, intOrder=11, lastExchangeTime=1532001260398, loc=false, ver=2.5.1#20180717-sha1:80e51c80, isClient=true] [2018-07-19 14:55:28,899][INFO ][disco-event-worker-#61] Topology snapshot [ver=27, servers=3, clients=4, CPUs=96, offheap=260.0GB, heap=56.0GB] [2018-07-19 14:55:28,899][INFO ][disco-event-worker-#61] Coordinator changed [prev=TcpDiscoveryNode [id=7240957f-a51b-452d-bfc8-420e8ef9ea68, addrs=[127.0.0.1, 172.17.0.1, 172.25.1.15], sockAddrs=[/172.17.0.1:0, /127.0.0.1:0, lab15.gridgain.local/172.25.1.15:0], discPort=0, order=16, intOrder=11, lastExchangeTime=1532001260398, loc=false, ver=2.5.1#20180717-sha1:80e51c80, isClient=true], cur=TcpDiscoveryNode [id=760fd8f2-b9d7-4953-aa86-3954c05c9feb, addrs=[127.0.0.1, 172.17.0.1, 172.25.1.21], sockAddrs=[/172.17.0.1:47500, lab21.gridgain.local/172.25.1.21:47500, /127.0.0.1:47500], discPort=47500, order=21, intOrder=15, lastExchangeTime=1532001260428, loc=false, ver=2.5.1#20180717-sha1:80e51c80, isClient=false]] [2018-07-19 14:55:28,899][INFO ][disco-event-worker-#61] ^-- Node [id=22B15E97-9944-48B5-A473-5C64E75A4D5A, clusterState=ACTIVE] [2018-07-19 14:55:28,899][INFO ][disco-event-worker-#61] ^-- Baseline [id=6, size=3, online=3, offline=0] [2018-07-19 14:55:28,899][INFO ][disco-event-worker-#61] Data Regions Configured: [2018-07-19 14:55:28,900][INFO ][disco-event-worker-#61] ^-- default [initSize=256.0 MiB, maxSize=60.0 GiB, persistenceEnabled=true] {noformat} The *Coordinator changed* message should not be here because in fact the coordinator was not changed. -- This message was sent by Atlassian JIRA (v7.6.3#76005)