[jira] [Commented] (IGNITE-3153) TcpDiscoveryZookeeperIpFinder doesn't properly handle client reconnections
[
https://issues.apache.org/jira/browse/IGNITE-3153?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15320679#comment-15320679
]
Valentin Kulichenko commented on IGNITE-3153:
-
Anton,
I was under impression that {{onSpiContextDestroyed}} can be called on client
disconnect. If this is not true, I think you're right. But the usage of
{{initGuard}} should be revisited in any case.
> TcpDiscoveryZookeeperIpFinder doesn't properly handle client reconnections
> --
>
> Key: IGNITE-3153
> URL: https://issues.apache.org/jira/browse/IGNITE-3153
> Project: Ignite
> Issue Type: Bug
> Components: general
>Affects Versions: 1.5.0.final
>Reporter: Valentin Kulichenko
>Assignee: Anton Vinogradov
> Fix For: 1.7
>
>
> The exception below is possible when client reconnects and ZooKeeper IP
> finder is used. Most likely this is caused by the fact that {{initGuard}} is
> flipped back to {{false}} when the context is destroyed and new curator
> instance is not created during the reconnect.
> This should be fixed and test coverage for this scenario should be improved.
> {noformat}
> 2016-05-16 12:00:59,096 5786 ERROR
> org.apache.ignite.spi.discovery.tcp.TcpDiscoverySpi Runtime error caught
> during grid runnable execution: IgniteSpiThread
> [name=tcp-client-disco-reconnector-#5%Default%]
> java.lang.IllegalStateException: Cannot be started more than once
> at
> org.apache.curator.framework.imps.CuratorFrameworkImpl.start(CuratorFrameworkImpl.java:237)
> ...
> at
> org.apache.ignite.spi.discovery.tcp.ipfinder.zk.TcpDiscoveryZookeeperIpFinder.init(TcpDiscoveryZookeeperIpFinder.java:144)
> at
> org.apache.ignite.spi.discovery.tcp.ipfinder.zk.TcpDiscoveryZookeeperIpFinder.getRegisteredAddresses(TcpDiscoveryZookeeperIpFinder.java:169)
> at
> org.apache.ignite.spi.discovery.tcp.TcpDiscoverySpi.registeredAddresses(TcpDiscoverySpi.java:1603)
> at
> org.apache.ignite.spi.discovery.tcp.TcpDiscoverySpi.resolvedAddresses(TcpDiscoverySpi.java:1552)
> at
> org.apache.ignite.spi.discovery.tcp.ClientImpl.joinTopology(ClientImpl.java:475)
> at
> org.apache.ignite.spi.discovery.tcp.ClientImpl.access$900(ClientImpl.java:118)
> at
> org.apache.ignite.spi.discovery.tcp.ClientImpl$Reconnector.body(ClientImpl.java:1175)
> at org.apache.ignite.spi.IgniteSpiThread.run(IgniteSpiThread.java:62)
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (IGNITE-3153) TcpDiscoveryZookeeperIpFinder doesn't properly handle client reconnections
[
https://issues.apache.org/jira/browse/IGNITE-3153?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15320535#comment-15320535
]
Anton Vinogradov commented on IGNITE-3153:
--
I wonder how can happen context destroy and then reconnect, except race at node
stop.
Seems that correct solution is to remove
if (!initGuard.compareAndSet(true, false))
return;
at onSpiContextDestroyed
This will fix problem, I think.
> TcpDiscoveryZookeeperIpFinder doesn't properly handle client reconnections
> --
>
> Key: IGNITE-3153
> URL: https://issues.apache.org/jira/browse/IGNITE-3153
> Project: Ignite
> Issue Type: Bug
> Components: general
>Affects Versions: 1.5.0.final
>Reporter: Valentin Kulichenko
>Assignee: Anton Vinogradov
> Fix For: 1.7
>
>
> The exception below is possible when client reconnects and ZooKeeper IP
> finder is used. Most likely this is caused by the fact that {{initGuard}} is
> flipped back to {{false}} when the context is destroyed and new curator
> instance is not created during the reconnect.
> This should be fixed and test coverage for this scenario should be improved.
> {noformat}
> 2016-05-16 12:00:59,096 5786 ERROR
> org.apache.ignite.spi.discovery.tcp.TcpDiscoverySpi Runtime error caught
> during grid runnable execution: IgniteSpiThread
> [name=tcp-client-disco-reconnector-#5%Default%]
> java.lang.IllegalStateException: Cannot be started more than once
> at
> org.apache.curator.framework.imps.CuratorFrameworkImpl.start(CuratorFrameworkImpl.java:237)
> ...
> at
> org.apache.ignite.spi.discovery.tcp.ipfinder.zk.TcpDiscoveryZookeeperIpFinder.init(TcpDiscoveryZookeeperIpFinder.java:144)
> at
> org.apache.ignite.spi.discovery.tcp.ipfinder.zk.TcpDiscoveryZookeeperIpFinder.getRegisteredAddresses(TcpDiscoveryZookeeperIpFinder.java:169)
> at
> org.apache.ignite.spi.discovery.tcp.TcpDiscoverySpi.registeredAddresses(TcpDiscoverySpi.java:1603)
> at
> org.apache.ignite.spi.discovery.tcp.TcpDiscoverySpi.resolvedAddresses(TcpDiscoverySpi.java:1552)
> at
> org.apache.ignite.spi.discovery.tcp.ClientImpl.joinTopology(ClientImpl.java:475)
> at
> org.apache.ignite.spi.discovery.tcp.ClientImpl.access$900(ClientImpl.java:118)
> at
> org.apache.ignite.spi.discovery.tcp.ClientImpl$Reconnector.body(ClientImpl.java:1175)
> at org.apache.ignite.spi.IgniteSpiThread.run(IgniteSpiThread.java:62)
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (IGNITE-3153) TcpDiscoveryZookeeperIpFinder doesn't properly handle client reconnections
[
https://issues.apache.org/jira/browse/IGNITE-3153?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15306883#comment-15306883
]
Valentin Kulichenko commented on IGNITE-3153:
-
Thanks, Roman! I added the trace.
> TcpDiscoveryZookeeperIpFinder doesn't properly handle client reconnections
> --
>
> Key: IGNITE-3153
> URL: https://issues.apache.org/jira/browse/IGNITE-3153
> Project: Ignite
> Issue Type: Bug
> Components: general
>Affects Versions: 1.5.0.final
>Reporter: Valentin Kulichenko
>Assignee: Valentin Kulichenko
> Fix For: 1.7
>
>
> The exception below is possible when client reconnects and ZooKeeper IP
> finder is used. Most likely this is caused by the fact that {{initGuard}} is
> flipped back to {{false}} when the context is destroyed and new curator
> instance is not created during the reconnect.
> This should be fixed and test coverage for this scenario should be improved.
> {noformat}
> 2016-05-16 12:00:59,096 5786 ERROR
> org.apache.ignite.spi.discovery.tcp.TcpDiscoverySpi Runtime error caught
> during grid runnable execution: IgniteSpiThread
> [name=tcp-client-disco-reconnector-#5%Default%]
> java.lang.IllegalStateException: Cannot be started more than once
> at
> org.apache.curator.framework.imps.CuratorFrameworkImpl.start(CuratorFrameworkImpl.java:237)
> ...
> at
> org.apache.ignite.spi.discovery.tcp.ipfinder.zk.TcpDiscoveryZookeeperIpFinder.init(TcpDiscoveryZookeeperIpFinder.java:144)
> at
> org.apache.ignite.spi.discovery.tcp.ipfinder.zk.TcpDiscoveryZookeeperIpFinder.getRegisteredAddresses(TcpDiscoveryZookeeperIpFinder.java:169)
> at
> org.apache.ignite.spi.discovery.tcp.TcpDiscoverySpi.registeredAddresses(TcpDiscoverySpi.java:1603)
> at
> org.apache.ignite.spi.discovery.tcp.TcpDiscoverySpi.resolvedAddresses(TcpDiscoverySpi.java:1552)
> at
> org.apache.ignite.spi.discovery.tcp.ClientImpl.joinTopology(ClientImpl.java:475)
> at
> org.apache.ignite.spi.discovery.tcp.ClientImpl.access$900(ClientImpl.java:118)
> at
> org.apache.ignite.spi.discovery.tcp.ClientImpl$Reconnector.body(ClientImpl.java:1175)
> at org.apache.ignite.spi.IgniteSpiThread.run(IgniteSpiThread.java:62)
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (IGNITE-3153) TcpDiscoveryZookeeperIpFinder doesn't properly handle client reconnections
[
https://issues.apache.org/jira/browse/IGNITE-3153?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15290508#comment-15290508
]
Roman Shtykh commented on IGNITE-3153:
--
[~vkulichenko] Seems you forgot to paste the details of the exception ;)
> TcpDiscoveryZookeeperIpFinder doesn't properly handle client reconnections
> --
>
> Key: IGNITE-3153
> URL: https://issues.apache.org/jira/browse/IGNITE-3153
> Project: Ignite
> Issue Type: Bug
> Components: general
>Affects Versions: 1.5.0.final
>Reporter: Valentin Kulichenko
> Fix For: 1.7
>
>
> The exception below is possible when client reconnects and ZooKeeper IP
> finder is used. Most likely this is caused by the fact that {{initGuard}} is
> flipped back to {{false}} when the context is destroyed and new curator
> instance is not created during the reconnect.
> This should be fixed and test coverage for this scenario should be improved.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
