You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@giraph.apache.org by Craig Muchinsky <cm...@us.ibm.com> on 2014/03/19 13:39:56 UTC
Request re-issues
Can anybody help me understand what underlying issue might be causing
these request re-issues a handful of times during graph execution? It
appears to be recovering, however there is a significant time delay
waiting for the request to time out, so I would rather understand the root
cause and fix it. The part that is curious to me is that all of the logged
fields (connected, future done, success, cause) seem to indicate that
everything worked fine.
2014-03-18 15:47:42,925 WARN org.apache.giraph.comm.netty.NettyClient:
checkRequestsForProblems: Problem with request id (destTask=148,reqId=24)
connected = true, future done = true, success = true, cause = null,
elapsed time = 602583, destination = host/1.1.1.1:30148
(reqId=24,destAddr=host:30148,elapsedNanos=602583065000,started=Tue Mar 18
15:37:40 EDT 2014,writeDone=true,writeSuccess=true)
2014-03-18 15:47:42,925 INFO org.apache.giraph.comm.netty.NettyClient:
checkRequestsForProblems: Re-issuing request
(reqId=24,destAddr=host:30148,elapsedNanos=27000,started=Tue Mar 18
15:47:42 EDT 2014)
2014-03-18 15:47:42,929 INFO
org.apache.giraph.comm.netty.handler.ResponseClientHandler:
messageReceived: Already completed request (taskId = 148, requestId = 24)
As a bit of background, I'm running with 432 cores, 216 workers, 2
input/compute threads, and 216 partitions, with data is being loaded using
both a vertex and edge input format.
Thanks,
Craig M.