ignite-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From vkulichenko <valentin.kuliche...@gmail.com>
Subject Re: One failing node stalling the whole cluster
Date Thu, 22 Sep 2016 22:46:19 GMT
Hi Sparkle,

Please properly subscribe to the mailing list so that the community can
receive email notifications for your messages. To subscribe, send empty
email to user-subscribe@ignite.apache.org and follow simple instructions in
the reply.

sparkle_j wrote
> In one of our tests, we noticed that Ignite's TcpCommunicationSpi object
> is growing and retaining over 70% of heap memory. Not sure why. Is this a
> known issue..? Please let us know if this is already addressed.
> We are testing with Ignite1.5, 10GB heap, with moderate amount of data on
> start up and just connect and disconnect dummy clients to grid. This is
> causing the heap to grow over time and eventually Full GC and node is
> being killed by Ignite. This is also causing cluster wide instability.
> Please see HeapDump screenshots attached.
> TcpCommSpi.png
> <http://apache-ignite-users.70518.x6.nabble.com/file/n7879/TcpCommSpi.png>  
> dom-tree-1.png
> <http://apache-ignite-users.70518.x6.nabble.com/file/n7879/dom-tree-1.png>  
> dom-tree.png
> <http://apache-ignite-users.70518.x6.nabble.com/file/n7879/dom-tree.png>  
Can you take the full heap dump and upload it somewhere?


View this message in context: http://apache-ignite-users.70518.x6.nabble.com/One-failing-node-stalling-the-whole-cluster-tp5372p7887.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.

View raw message