You can try to kill them, but i suppose: you can`t, cause specific state:
MARKED_ROLLBACK just try.
If this can`t help — you need to grep all ignite.log(s) and found this xid=
versions, i suppose you will grep possibly pair instances, after you need to
sequentially restart grid instances with such xid in logs. If all ok — just
plan to upgrade into 2.10, there are some bugs on 2.8.1 closing to your case.
thanks!
>Hi,
>thanks for the hint using the control.sh script. It looks like the server node
>with .111 IP address is having some very old transactions:
>Matching transactions:
>TcpDiscoveryNode [id=78b6df9f-7b77-46ea-9aef-a1b61918b258,
>addrs=[110.10.123.111], order=1, ver=2.8.1#20200521-sha1:86422096,
>isClient=false, consistentId=3d51530f-0dec-46da-98cb-eb550c860777]
> Tx: [xid=93724cf8971-00000000-0d9e-4326-0000-000000000001, label=null,
>state=MARKED_ROLLBACK, startTime=2021-05-24 06:22:10.778, duration=3224128,
>isolation=REPEATABLE_READ, concurrency=PESSIMISTIC,
>topVer=AffinityTopologyVersion [topVer=74, minorTopVer=0], timeout=9990,
>size=0, dhtNodes=[], nearXid=f0724cf8971-00000000-0d9e-4326-0000-00000000004a,
>parentNodeIds=[75431074]]
> Tx: [xid=7219dcbf971-00000000-0d9e-452c-0000-000000000001, label=null,
>state=MARKED_ROLLBACK, startTime=2021-06-23 13:12:53.094, duration=607486,
>isolation=REPEATABLE_READ, concurrency=PESSIMISTIC,
>topVer=AffinityTopologyVersion [topVer=592, minorTopVer=0], timeout=10000,
>size=0, dhtNodes=[], nearXid=f119dcbf971-00000000-0d9e-452c-0000-0000000001d9,
>parentNodeIds=[458bca81]]
> Tx: [xid=a1480fb5a71-00000000-0d9e-4593-0000-000000000001, label=null,
>state=PREPARING, startTime=2021-06-30 10:51:37.874, duration=11161,
>isolation=READ_COMMITTED, concurrency=OPTIMISTIC,
>topVer=AffinityTopologyVersion [topVer=695, minorTopVer=0], timeout=0, size=1,
>dhtNodes=[78b6df9f], nearXid=a1480fb5a71-00000000-0d9e-4593-0000-000000000001,
>parentNodeIds=[78b6df9f]]
>TcpDiscoveryNode [id=5eb48408-a239-4a92-9fb2-7903407975ad,
>addrs=[110.10.123.87], order=688, ver=2.8.1#20200521-sha1:86422096,
>isClient=true, consistentId=5eb48408-a239-4a92-9fb2-7903407975ad]
> Tx: [xid=e95d3fb5a71-00000000-0d9e-4593-0000-0000000002b0, label=null,
>state=PREPARING, startTime=2021-06-30 13:57:39.511, duration=0,
>isolation=REPEATABLE_READ, concurrency=PESSIMISTIC,
>topVer=AffinityTopologyVersion [topVer=695, minorTopVer=0], timeout=0, size=1,
>dhtNodes=[78b6df9f], nearXid=e95d3fb5a71-00000000-0d9e-4593-0000-0000000002b0,
>parentNodeIds=[5eb48408]]
>Command [TX] finished with code: 0
>What happens if I would kill these two transactions that are in
>MARKED_ROLLBACK state but look like they never finished? Is it safe to kill
>them?
>Thanks!
>
>On 30.06.21 15:20, Zhenya Stanilovsky wrote:
>>
>>Hi, first of all you need to know who is holding a lock, check [1] command.
>>
>>[1]
>>https://ignite.apache.org/docs/latest/tools/control-script#transaction-management
>>
>>>Hi all,
>>>
>>>I have the situation that at least one DB row of my persistent cache
>>>seems locked and I can't change it anymore. Everytime I want to change
>>>it using SQL a TransactionTimeoutException happens like this:
>>>
>>>class org.apache.ignite.transactions.TransactionTimeoutException: Failed
>>>to acquire lock within provided timeout for transaction [timeout=10000,
>>>tx=GridNearTxLocal[xid=cc2d4db5a71-00000000-0d9e-4593-0000-0000000002b6,
>>>xidVersion=GridCacheVersion [topVer=228476307, order=1625038312140,
>>>nodeOrder=694], nearXidVersion=GridCacheVersion [topVer=228476307,
>>>order=1625038312140, nodeOrder=694], concurrency=PESSIMISTIC,
>>>isolation=REPEATABLE_READ, state=MARKED_ROLLBACK, invalidate=false,
>>>rollbackOnly=true, nodeId=62e91173-e912-49d5-b238-e95f8fe38314,
>>>timeout=10000, startTime=1625051540461, duration=10034, label=null]]
>>>
>>>However, the system continues to run and other objects in the cache can
>>>be added and changed.
>>>
>>>Is there a way to unlock this particular entry?
>>>
>>>Thanks!
>>>
>>
>>
>>
>>
