[jira] [Commented] (IGNITE-12636) Full rebalance instead of historical one
[
https://issues.apache.org/jira/browse/IGNITE-12636?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17032176#comment-17032176
]
Mirza Aliev commented on IGNITE-12636:
--
The problem with ZooKeeper test doesn't seem to be connected with the issue, so
I would say that visa is received.
> Full rebalance instead of historical one
>
>
> Key: IGNITE-12636
> URL: https://issues.apache.org/jira/browse/IGNITE-12636
> Project: Ignite
> Issue Type: Bug
>Reporter: Mirza Aliev
>Assignee: Mirza Aliev
>Priority: Major
> Time Spent: 10m
> Remaining Estimate: 0h
>
> Common configuration:
> 1)4 nodes
> 2)3-4 caches
> 3)IGNITE_PDS_WAL_REBALANCE_THRESHOLD=500
> 4)walHistorySize=500
> 5)IGNITE_PDS_MAX_CHECKPOINT_MEMORY_HISTORY_SIZE=500
> Scenario:
> 1)Load a lot of data.
> 2)Start the transaction on some client but DON'T close it.
> 3)Stop the server node.
> 4)Start the server.
> 5)PME should be started
> 6)Kill the client started the transaction
> Expected result:
> Rebalance is HISTORICAL.
> Actual result:
> Rebalance is full:
>
> {noformat}
> [12:01:58,613][INFO]sys-#95[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_6, supplier=5462dc46-f71f-49d8-8a1d-d9d69c3e372a,
> topic=0, fullPartitions=[23], histPartitions=[]]
> [12:01:58,614][INFO]sys-#109[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_6, supplier=8ab78982-0bcf-494f-a634-f3fb2d78328f,
> topic=0, fullPartitions=[1], histPartitions=[]]
> [12:01:58,614][INFO]sys-#101[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_6, supplier=8ab78982-0bcf-494f-a634-f3fb2d78328f,
> topic=1, fullPartitions=[55], histPartitions=[]]
> [12:01:59,004][INFO]sys-#99[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_4_118, supplier=5462dc46-f71f-49d8-8a1d-d9d69c3e372a,
> topic=0, fullPartitions=[5], histPartitions=[]]
> [12:01:59,004][INFO]sys-#96[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_4_118, supplier=48e2a2b5-2119-4b5c-873c-eb8d0c436b6a,
> topic=0, fullPartitions=[15], histPartitions=[]]
> [12:01:59,196][INFO]sys-#104[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_2_058, supplier=48e2a2b5-2119-4b5c-873c-eb8d0c436b6a,
> topic=0, fullPartitions=[21], histPartitions=[]]
> [12:01:59,196][INFO]sys-#95[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_2_058, supplier=8ab78982-0bcf-494f-a634-f3fb2d78328f,
> topic=0, fullPartitions=[19], histPartitions=[]]{noformat}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
[jira] [Commented] (IGNITE-12636) Full rebalance instead of historical one
[
https://issues.apache.org/jira/browse/IGNITE-12636?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17032175#comment-17032175
]
Ignite TC Bot commented on IGNITE-12636:
{panel:title=Branch: [pull/7379/head] Base: [master] : Possible Blockers
(1)|borderStyle=dashed|borderColor=#ccc|titleBGColor=#F7D6C1}
{color:#d04437}ZooKeeper (Discovery) 2{color} [[tests 0 TIMEOUT , Exit Code
|https://ci.ignite.apache.org/viewLog.html?buildId=4992018]]
{panel}
[TeamCity *-- Run :: All*
Results|https://ci.ignite.apache.org/viewLog.html?buildId=4991097buildTypeId=IgniteTests24Java8_RunAll]
> Full rebalance instead of historical one
>
>
> Key: IGNITE-12636
> URL: https://issues.apache.org/jira/browse/IGNITE-12636
> Project: Ignite
> Issue Type: Bug
>Reporter: Mirza Aliev
>Assignee: Mirza Aliev
>Priority: Major
> Time Spent: 10m
> Remaining Estimate: 0h
>
> Common configuration:
> 1)4 nodes
> 2)3-4 caches
> 3)IGNITE_PDS_WAL_REBALANCE_THRESHOLD=500
> 4)walHistorySize=500
> 5)IGNITE_PDS_MAX_CHECKPOINT_MEMORY_HISTORY_SIZE=500
> Scenario:
> 1)Load a lot of data.
> 2)Start the transaction on some client but DON'T close it.
> 3)Stop the server node.
> 4)Start the server.
> 5)PME should be started
> 6)Kill the client started the transaction
> Expected result:
> Rebalance is HISTORICAL.
> Actual result:
> Rebalance is full:
>
> {noformat}
> [12:01:58,613][INFO]sys-#95[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_6, supplier=5462dc46-f71f-49d8-8a1d-d9d69c3e372a,
> topic=0, fullPartitions=[23], histPartitions=[]]
> [12:01:58,614][INFO]sys-#109[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_6, supplier=8ab78982-0bcf-494f-a634-f3fb2d78328f,
> topic=0, fullPartitions=[1], histPartitions=[]]
> [12:01:58,614][INFO]sys-#101[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_6, supplier=8ab78982-0bcf-494f-a634-f3fb2d78328f,
> topic=1, fullPartitions=[55], histPartitions=[]]
> [12:01:59,004][INFO]sys-#99[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_4_118, supplier=5462dc46-f71f-49d8-8a1d-d9d69c3e372a,
> topic=0, fullPartitions=[5], histPartitions=[]]
> [12:01:59,004][INFO]sys-#96[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_4_118, supplier=48e2a2b5-2119-4b5c-873c-eb8d0c436b6a,
> topic=0, fullPartitions=[15], histPartitions=[]]
> [12:01:59,196][INFO]sys-#104[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_2_058, supplier=48e2a2b5-2119-4b5c-873c-eb8d0c436b6a,
> topic=0, fullPartitions=[21], histPartitions=[]]
> [12:01:59,196][INFO]sys-#95[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_2_058, supplier=8ab78982-0bcf-494f-a634-f3fb2d78328f,
> topic=0, fullPartitions=[19], histPartitions=[]]{noformat}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
[jira] [Commented] (IGNITE-12636) Full rebalance instead of historical one
[
https://issues.apache.org/jira/browse/IGNITE-12636?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17031646#comment-17031646
]
Mirza Aliev commented on IGNITE-12636:
--
We reproduced the problem and found how to fix that. Long story short, problem
can be reproduced when client node stops during PME that triggers rebalance.
There is no need to start transaction, that was described above.
Root cause is the logic with GridDhtPreloader.generateAssignments(), we do not
need to generateAssignments for client exchange task
> Full rebalance instead of historical one
>
>
> Key: IGNITE-12636
> URL: https://issues.apache.org/jira/browse/IGNITE-12636
> Project: Ignite
> Issue Type: Bug
>Reporter: Mirza Aliev
>Assignee: Mirza Aliev
>Priority: Major
>
> Common configuration:
> 1)4 nodes
> 2)3-4 caches
> 3)IGNITE_PDS_WAL_REBALANCE_THRESHOLD=500
> 4)walHistorySize=500
> 5)IGNITE_PDS_MAX_CHECKPOINT_MEMORY_HISTORY_SIZE=500
> Scenario:
> 1)Load a lot of data.
> 2)Start the transaction on some client but DON'T close it.
> 3)Stop the server node.
> 4)Start the server.
> 5)PME should be started
> 6)Kill the client started the transaction
> Expected result:
> Rebalance is HISTORICAL.
> Actual result:
> Rebalance is full:
>
> {noformat}
> [12:01:58,613][INFO]sys-#95[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_6, supplier=5462dc46-f71f-49d8-8a1d-d9d69c3e372a,
> topic=0, fullPartitions=[23], histPartitions=[]]
> [12:01:58,614][INFO]sys-#109[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_6, supplier=8ab78982-0bcf-494f-a634-f3fb2d78328f,
> topic=0, fullPartitions=[1], histPartitions=[]]
> [12:01:58,614][INFO]sys-#101[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_6, supplier=8ab78982-0bcf-494f-a634-f3fb2d78328f,
> topic=1, fullPartitions=[55], histPartitions=[]]
> [12:01:59,004][INFO]sys-#99[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_4_118, supplier=5462dc46-f71f-49d8-8a1d-d9d69c3e372a,
> topic=0, fullPartitions=[5], histPartitions=[]]
> [12:01:59,004][INFO]sys-#96[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_4_118, supplier=48e2a2b5-2119-4b5c-873c-eb8d0c436b6a,
> topic=0, fullPartitions=[15], histPartitions=[]]
> [12:01:59,196][INFO]sys-#104[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_2_058, supplier=48e2a2b5-2119-4b5c-873c-eb8d0c436b6a,
> topic=0, fullPartitions=[21], histPartitions=[]]
> [12:01:59,196][INFO]sys-#95[GridDhtPartitionDemander] Started rebalance
> routine [cache_group_2_058, supplier=8ab78982-0bcf-494f-a634-f3fb2d78328f,
> topic=0, fullPartitions=[19], histPartitions=[]]{noformat}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
