[jira] [Commented] (IGNITE-4937) Transactions deadlock detection should happen on each grid hang
[ https://issues.apache.org/jira/browse/IGNITE-4937?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16034513#comment-16034513 ] Anton Vinogradov commented on IGNITE-4937: -- [~avinogradov] Just test comment > Transactions deadlock detection should happen on each grid hang > --- > > Key: IGNITE-4937 > URL: https://issues.apache.org/jira/browse/IGNITE-4937 > Project: Ignite > Issue Type: Task > Components: cache >Reporter: Anton Vinogradov >Assignee: Anton Vinogradov >Priority: Critical > Fix For: 2.1 > > > Each "Failed to wait for partition map exchange" or similar should cause such > check. > Check can be added to > org.apache.ignite.internal.processors.cache.GridCachePartitionExchangeManager#dumpPendingObjects > or/and > org.apache.ignite.internal.processors.cache.GridCachePartitionExchangeManager#dumpLongRunningOperations -- This message was sent by Atlassian JIRA (v6.3.15#6346)
[jira] [Commented] (IGNITE-4937) Transactions deadlock detection should happen on each grid hang
[ https://issues.apache.org/jira/browse/IGNITE-4937?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15976892#comment-15976892 ] Anton Vinogradov commented on IGNITE-4937: -- One more idea is to set default timeout not equals to infinity, for example - 1 minute. And provide availability to set another value of this timeout. > Transactions deadlock detection should happen on each grid hang > --- > > Key: IGNITE-4937 > URL: https://issues.apache.org/jira/browse/IGNITE-4937 > Project: Ignite > Issue Type: Task >Reporter: Anton Vinogradov >Assignee: Anton Vinogradov >Priority: Critical > Fix For: 2.1 > > > Each "Failed to wait for partition map exchange" or similar should cause such > check. > Check can be added to > org.apache.ignite.internal.processors.cache.GridCachePartitionExchangeManager#dumpPendingObjects > or/and > org.apache.ignite.internal.processors.cache.GridCachePartitionExchangeManager#dumpLongRunningOperations -- This message was sent by Atlassian JIRA (v6.3.15#6346)
[jira] [Commented] (IGNITE-4937) Transactions deadlock detection should happen on each grid hang
[ https://issues.apache.org/jira/browse/IGNITE-4937?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15962837#comment-15962837 ] Andrey Gura commented on IGNITE-4937: - >From my point of view it is helpful improvement but we should be careful here >because we can block local operation by starting distributed operation. Also I think we should add deadlock detection triggers that can be activated periodically (configuration setting) or manually (e.g. via MBean). > Transactions deadlock detection should happen on each grid hang > --- > > Key: IGNITE-4937 > URL: https://issues.apache.org/jira/browse/IGNITE-4937 > Project: Ignite > Issue Type: Task >Reporter: Anton Vinogradov >Priority: Critical > > Each "Failed to wait for partition map exchange" or similar should cause such > check. > Check can be added to > org.apache.ignite.internal.processors.cache.GridCachePartitionExchangeManager#dumpPendingObjects > or/and > org.apache.ignite.internal.processors.cache.GridCachePartitionExchangeManager#dumpLongRunningOperations -- This message was sent by Atlassian JIRA (v6.3.15#6346)