Namespace should handle request rejected messages.

[marmbrus]

Instead of throwing this error it would be nice if it retried or something.

WAR [20101020-00:50:47.989] perf: java.lang.RuntimeException: java.lang.RuntimeException: Unexpected Message (was expecting either GetRangeResponse or GetPrefixResponse): {"reason": "Thread Pool Full", "req": {"minKey": {"bytes": "�User26682????�"}, "maxKey": {"bytes": "�User26682????�"}, "limit": 10, "offset": null, "ascending": false}}
WAR [20101020-00:50:47.989] perf: at edu.berkeley.cs.scads.comm.ComputationFuture$class.edu$berkeley$cs$scads$comm$ComputationFuture$$dispatchResult(Futures.scala:208)
WAR [20101020-00:50:47.989] perf: at edu.berkeley.cs.scads.comm.ComputationFuture$class.get(Futures.scala:218)
WAR [20101020-00:50:47.989] perf: at edu.berkeley.cs.scads.storage.QuorumProtocol$$anon$2.get(QuorumProtocol.scala:288)
WAR [20101020-00:50:47.989] perf: at edu.berkeley.cs.scads.comm.ComputationFuture$class.get(Futures.scala:204)
WAR [20101020-00:50:47.989] perf: at edu.berkeley.cs.scads.storage.QuorumProtocol$$anon$2.get(QuorumProtocol.scala:288)
WAR [20101020-00:50:47.989] perf: at edu.berkeley.cs.scads.piql.ParallelExecutor$$anon$10.edu$berkeley$cs$scads$piql$ParallelExecutor$$anon$$fillBuffer(Iterators.scala:600)
WAR [20101020-00:50:47.989] perf: at edu.berkeley.cs.scads.piql.ParallelExecutor$$anon$10$$anonfun$hasNext$1.apply$mcVI$sp(Iterators.scala:570)
WAR [20101020-00:50:47.989] perf: at scala.collection.immutable.Range$ByOne$class.foreach$mVc$sp(Range.scala:282)
WAR [20101020-00:50:47.989] perf: at scala.collection.immutable.Range$$anon$1.foreach$mVc$sp(Range.scala:274)
WAR [20101020-00:50:47.989] perf: at edu.berkeley.cs.scads.piql.ParallelExecutor$$anon$10.hasNext(Iterators.scala:568)
WAR [20101020-00:50:47.989] perf: at edu.berkeley.cs.scads.piql.SimpleExecutor$$anon$6.hasNext(Iterators.scala:369)
WAR [20101020-00:50:47.989] perf: at scala.collection.Iterator$class.foreach(Iterator.scala:631)
WAR [20101020-00:50:47.989] perf: at edu.berkeley.cs.scads.piql.QueryIterator.foreach(Iterators.scala:14)
WAR [20101020-00:50:47.989] perf: at scala.collection.generic.Growable$class.$plus$plus$eq(Growable.scala:48)
WAR [20101020-00:50:47.989] perf: at scala.collection.TraversableOnce$class.toList(TraversableOnce.scala:399)
WAR [20101020-00:50:47.989] perf: at edu.berkeley.cs.scads.piql.QueryIterator.toList(Iterators.scala:14)
WAR [20101020-00:50:47.989] perf: at edu.berkeley.cs.scads.piql.ScadrClient.exec(ScadrClient.scala:42)
WAR [20101020-00:50:47.989] perf: at edu.berkeley.cs.scads.piql.ScadrClient.thoughtstream(ScadrClient.scala:101)
WAR [20101020-00:50:47.989] perf: at edu.berkeley.cs.scads.perf.scadr.cardinality.LoadClient$$anonfun$run$1$$anonfun$apply$mcII$sp$1.apply(ScadrCardinality.scala:107)
WAR [20101020-00:50:47.989] perf: at edu.berkeley.cs.scads.perf.scadr.cardinality.LoadClient$$anonfun$run$1$$anonfun$apply$mcII$sp$1.apply(ScadrCardinality.scala:96)
WAR [20101020-00:50:47.989] perf: at deploylib.ParallelSeq$$anonfun$pmap$1$$anonfun$apply$3.apply(Concurrent.scala:77)
WAR [20101020-00:50:47.989] perf: at deploylib.Future.deploylib$Future$$tryCatch(Concurrent.scala:44)
WAR [20101020-00:50:47.989] perf: at deploylib.Future$$anon$1.run(Concurrent.scala:13)
WAR [20101020-00:50:47.989] perf: Caused by java.lang.RuntimeException: Unexpected Message (was expecting either GetRangeResponse or

[stephentu]

it should probably retry after waiting a while though- if a thread pool is full its usually a sign that the server is overloaded, so continuing to hammer it with requests probably is not a good idea.

[marmbrus]

agreed, tim and i talked about and think the right thing to do is either nothing, if we can still achieve quorum. Otherwise throw timeout or something.

Leave retry to whatever is above us.