ignite-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From vdpyatkov <vldpyat...@gmail.com>
Subject Re: what does remaps mean in DataStreamerImpl.java
Date Thu, 04 May 2017 19:50:20 GMT
Hi,

I have see that your grid suffer from long running operations:

[19:20:19,285][WARNING][grid-timeout-worker-#63%null%][GridCachePartitionExchangeManager]
Found long running cache future [startTime=19:18:55.302,
curTime=19:20:19.280, fut=GridDhtAtomicSingleUpdateFuture
[key=MyProject.TokenSecondMap [idHash=1526386777, hash=2035509954,
processingSecond=1174038983, token=38888], nearReaderEntry=null]]
[19:20:19,286][WARNING][grid-timeout-worker-#63%null%][GridCachePartitionExchangeManager]
Found long running cache future [startTime=19:18:55.302,
curTime=19:20:19.280,
 ...

by that reason it can not changing topology fast.

This is topology before killing nodes:

[19:17:42,341][INFO][disco-event-worker-#100%null%][GridDiscoveryManager]
Topology snapshot [ver=30, servers=30, clients=0, CPUs=168, heap=120.0GB]

and the same minor topology version after remaping failed out:

[19:20:41,302][WARNING][exchange-worker-#103%null%][GridCachePartitionExchangeManager]
Ready affinity version: AffinityTopologyVersion [topVer=30, minorTopVer=10]

But actual exchange future have required 33 version already:

[topVer=AffinityTopologyVersion [topVer=33, minorTopVer=0], nodeId=7484419b,
evt=NODE_FAILED]

Remap was failed, because grid had not been in time to change topology.

You should avoid message about long running operations (reduce the load or
change logic), otherwise exchange will going slow and you got the exception.
Also, long exchange can hang whole grid, not only lead to the streamer
exception.



--
View this message in context: http://apache-ignite-users.70518.x6.nabble.com/what-does-remaps-mean-in-DataStreamerImpl-java-tp12033p12440.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.

Mime
View raw message