[ https://issues.apache.org/jira/browse/IGNITE-9738?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16651729#comment-16651729 ]
Ivan Daschinskiy commented on IGNITE-9738: ------------------------------------------ [~v.pyatkov] generally, it looks good for me, but there are some strange failures in TC. Could you please rerun failed suites again? > Client node can suddenly fail on start > -------------------------------------- > > Key: IGNITE-9738 > URL: https://issues.apache.org/jira/browse/IGNITE-9738 > Project: Ignite > Issue Type: Bug > Reporter: Vladislav Pyatkov > Assignee: Vladislav Pyatkov > Priority: Major > > If client joining to large topology it can to spend some time on waiting > {{TcpDiscoveryNodeAddFinishedMessage}}, but in that time it can not to send > {{TcpDiscoveryClientMetricsUpdateMessage.}} By that reason server can to > reset client from topology. > {noformat} > Client node considered as unreachable and will be dropped from cluster, > because no metrics update messages received in interval: > TcpDiscoverySpi.clientFailureDetectionTimeout() ms. It may be caused by > network problems or long GC pause on client node, try to increase this > parameter. [nodeId=a3493895-7c13-403c-bf0d-94eab0000011, > clientFailureDetectionTimeout=30000]; > {noformat} > We should to sent {{TcpDiscoveryClientMetricsUpdateMessage as soon as > possible}} without, waiting finish of join procedure. -- This message was sent by Atlassian JIRA (v7.6.3#76005)