[jira] [Created] (IGNITE-5757) RENT partition state is set before exchange is completed

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

[jira] [Created] (IGNITE-5757) RENT partition state is set before exchange is completed

Anton Vinogradov (Jira)
Alexey Goncharuk created IGNITE-5757:
----------------------------------------

             Summary: RENT partition state is set before exchange is completed
                 Key: IGNITE-5757
                 URL: https://issues.apache.org/jira/browse/IGNITE-5757
             Project: Ignite
          Issue Type: Bug
          Components: cache
    Affects Versions: 2.1
            Reporter: Alexey Goncharuk
            Assignee: Alexey Goncharuk
             Fix For: 2.2


I observe the following error on TC during GridCachePartitionedGetAndTransformStoreSelfTest run:
{code}
[2017-07-14 16:26:49,857][ERROR][sys-stripe-3-#4%near.GridCachePartitionedGetAndTransformStoreSelfTest%][IgniteTxHandler] Failed to prepare DHT transaction: GridDhtTxLocal [nearNodeId=8a4cdb3f-85be-4286-a63d-728a6c800001, nearFutId=a8ef7414d51-e91dff67-b46c-4762-b7b9-9bcf382a82a0, nearMiniId=1, nearFinFutId=null, nearFinMiniId=0, nearXidVer=GridCacheVersion [topVer=111518810, order=1500038810278, nodeOrder=3], super=GridDhtTxLocalAdapter [nearOnOriginatingNode=true, nearNodes=[], dhtNodes=[], explicitLock=false, super=IgniteTxLocalAdapter [completedBase=null, sndTransformedVals=false, depEnabled=false, txState=IgniteTxImplicitSingleStateImpl [init=true, recovery=false], super=IgniteTxAdapter [xidVer=GridCacheVersion [topVer=111518810, order=1500038810305, nodeOrder=1], writeVer=null, implicit=true, loc=true, threadId=239, startTime=1500038809744, nodeId=322e96c3-eb1c-4aea-b9fd-3a3dae368d31, startVer=GridCacheVersion [topVer=111518810, order=1500038810305, nodeOrder=1], endVer=null, isolation=READ_COMMITTED, concurrency=OPTIMISTIC, timeout=0, sysInvalidate=false, sys=false, plc=2, commitVer=GridCacheVersion [topVer=111518810, order=1500038810305, nodeOrder=1], finalizing=NONE, invalidParts=null, state=ROLLED_BACK, timedOut=false, topVer=AffinityTopologyVersion [topVer=4, minorTopVer=0], duration=107ms, onePhaseCommit=false], size=0]]]
class org.apache.ignite.internal.transactions.IgniteTxRollbackCheckedException: Failed to prepare transaction: GridDhtTxLocal [nearNodeId=8a4cdb3f-85be-4286-a63d-728a6c800001, nearFutId=a8ef7414d51-e91dff67-b46c-4762-b7b9-9bcf382a82a0, nearMiniId=1, nearFinFutId=null, nearFinMiniId=0, nearXidVer=GridCacheVersion [topVer=111518810, order=1500038810278, nodeOrder=3], super=GridDhtTxLocalAdapter [nearOnOriginatingNode=true, nearNodes=[], dhtNodes=[], explicitLock=false, super=IgniteTxLocalAdapter [completedBase=null, sndTransformedVals=false, depEnabled=false, txState=IgniteTxImplicitSingleStateImpl [init=true, recovery=false], super=IgniteTxAdapter [xidVer=GridCacheVersion [topVer=111518810, order=1500038810305, nodeOrder=1], writeVer=null, implicit=true, loc=true, threadId=239, startTime=1500038809744, nodeId=322e96c3-eb1c-4aea-b9fd-3a3dae368d31, startVer=GridCacheVersion [topVer=111518810, order=1500038810305, nodeOrder=1], endVer=null, isolation=READ_COMMITTED, concurrency=OPTIMISTIC, timeout=0, sysInvalidate=false, sys=false, plc=2, commitVer=null, finalizing=NONE, invalidParts=null, state=MARKED_ROLLBACK, timedOut=false, topVer=AffinityTopologyVersion [topVer=4, minorTopVer=0], duration=97ms, onePhaseCommit=false], size=0]]]
        at org.apache.ignite.internal.processors.cache.distributed.dht.GridDhtTxLocal.prepareAsync(GridDhtTxLocal.java:418)
        at org.apache.ignite.internal.processors.cache.transactions.IgniteTxHandler.prepareNearTx(IgniteTxHandler.java:459)
        at org.apache.ignite.internal.processors.cache.transactions.IgniteTxHandler.processNearTxPrepareRequest(IgniteTxHandler.java:121)
        at org.apache.ignite.internal.processors.cache.transactions.IgniteTxHandler.access$000(IgniteTxHandler.java:95)
        at org.apache.ignite.internal.processors.cache.transactions.IgniteTxHandler$1.apply(IgniteTxHandler.java:141)
        at org.apache.ignite.internal.processors.cache.transactions.IgniteTxHandler$1.apply(IgniteTxHandler.java:139)
        at org.apache.ignite.internal.processors.cache.GridCacheIoManager.processMessage(GridCacheIoManager.java:1042)
        at org.apache.ignite.internal.processors.cache.GridCacheIoManager.onMessage0(GridCacheIoManager.java:561)
        at org.apache.ignite.internal.processors.cache.GridCacheIoManager.handleMessage(GridCacheIoManager.java:378)
        at org.apache.ignite.internal.processors.cache.GridCacheIoManager.handleMessage(GridCacheIoManager.java:304)
        at org.apache.ignite.internal.processors.cache.GridCacheIoManager.access$100(GridCacheIoManager.java:99)
        at org.apache.ignite.internal.processors.cache.GridCacheIoManager$1.onMessage(GridCacheIoManager.java:293)
        at org.apache.ignite.internal.managers.communication.GridIoManager.invokeListener(GridIoManager.java:1556)
        at org.apache.ignite.internal.managers.communication.GridIoManager.processRegularMessage0(GridIoManager.java:1184)
        at org.apache.ignite.internal.managers.communication.GridIoManager.access$4200(GridIoManager.java:126)
        at org.apache.ignite.internal.managers.communication.GridIoManager$9.run(GridIoManager.java:1097)
        at org.apache.ignite.internal.util.StripedExecutor$Stripe.run(StripedExecutor.java:483)
        at java.lang.Thread.run(Thread.java:745)
Caused by: class org.apache.ignite.IgniteCheckedException: Adding entry to partition that is concurrently evicted [part=58, shouldBeMoving=false]
        at org.apache.ignite.internal.processors.cache.distributed.dht.GridDhtTxLocalAdapter.addEntry(GridDhtTxLocalAdapter.java:523)
        at org.apache.ignite.internal.processors.cache.distributed.dht.GridDhtTxLocal.prepareAsync(GridDhtTxLocal.java:399)
        ... 17 more
{code}

The issue happens because GridDhtLocalTransaction is being prepared on topology version (4, 0), and the next topology version (4, 1) changes the primary node assignment. The topology is locked on the originating node, which is a server node, at the same time topology is not locked on primary node that is about to rent it's partitions.
When exchange runs on primary node, it does not see the GridDhtTxLocal transaction yet, completes the partition release future and rents all partitions.
After this, the GridDhtTxLocat starts prepare phase and gets the exception above.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)