Error : Timeout while delivering a packet

Question

Hi,
When we handle objects in the cache, we get the following error (after several hundreds operations) :
2005-12-15 07:55:37.551 Tangosol Coherence 3.1/321 <Warning> (thread=PacketPublisher, member=1): Timeout while delivering a packet; requesting the departure confirmation for Member(Id=2, Timestamp=Thu Dec 15 07:53:47 CET 2005, Address=192.168.1.61, Port=8088, MachineId=26941)
by MemberSet(Size=1, BitSetCount=1
Member(Id=3, Timestamp=Thu Dec 15 07:53:47 CET 2005, Address=192.168.1.63, Port=8088, MachineId=26943)
2005-12-15 07:55:37.554 Tangosol Coherence 3.1/321 <Info> (thread=Cluster, member=1): Member departure confirmed; removing Member(Id=2, Timestamp=Thu Dec 15 07:53:47 CET 2005, Address=192.168.1.61, Port=8088, MachineId=26941)
2005-12-15 07:55:37.592 Tangosol Coherence 3.1/321 <Warning> (thread=DistributedCache, member=1): Assigned 85 orphaned primary buckets
Our configuration is a partitioned (distributed) cache between 3 nodes, each on a specific server :
- 192.168.1.61;
- 192.168.1.62; (we get the error message on this node)
- 192.168.1.63;
Here are our two XML files config.
Note that we use specific WKA, because multicast is blocked by the switch between the 3 servers.
Coherence is launched by a web app, in JOnAS 4.5.3 (JVM 1.5.0). The aim is not to cache sessions, but some data like in a classic application.
I attach the trace, from node 2, where you can see that some operations take more than 50 sec to be performed! It is just before the node problem.
We use transaction to avoid incoherency between concurrent access.
Thanks in advance,
Vincent<br><br> <b> Attachment: </b><br>coherence-cache-config.xml <br> (*To use this attachment you will need to rename 206.bin to coherence-cache-config.xml after the download is complete.)<br><br> <b> Attachment: </b><br>tangosol-coherence.xml <br> (*To use this attachment you will need to rename 207.bin to tangosol-coherence.xml after the download is complete.)<br><br> <b> Attachment: </b><br>trace_s2_end.txt <br> (*To use this attachment you will need to rename 208.bin to trace_s2_end.txt after the download is complete.)

Answer 1

Hi Vincent,
When this happens, please email all 3 nodes logs (with -Dtangosol.coherence.log.level=5 log level) and thread dumps (pereferably several, taken at some time intervals) to support at tangosol.com.
Regards,
Dimitri