I'm having issues with ManifoldCF losing connection to ZooKeeper. This is easily repeatable: I just need to leave ManifoldCF running for a few days. The results are not always "No route to host" as I previously reported -- sometimes its just connect timeouts or other behavior, but the connection is always lost after a few days. I have several other projects running in the same environment connecting to ZooKeeper and none of them experience any issues at all, so this is something specific to MCF and not the environment or ZK.
Is there any way I can at least debug this so I can provide better information on which to file an issue? Thanks, Raman On Mon, Apr 12, 2021 at 3:33 PM Raman Gupta <rocketra...@gmail.com> wrote: > After my agent process has been up for a while, it starts producing these > logs repeatedly: > > java.net.NoRouteToHostException: No route to host > at java.base/sun.nio.ch.SocketChannelImpl.checkConnect(Native Method) > at java.base/sun.nio.ch.SocketChannelImpl.finishConnect(SocketChannelIm > at org.apache.zookeeper.ClientCnxnSocketNIO.doTransport(ClientCnxnSocke > at org.apache.zookeeper.ClientCnxn$SendThread.run(ClientCnxn.java:1141) > > The ZooKeeper cluster itself is up and responding normally, and the > network is available, and the zookeeper host resolves, though a transient > lookup failure cannot be ruled out, as ManifoldCF and Zookeeper are running > inside Kubernetes. A restart of the agent promptly fixes the issue. > > Regards, > Raman > >
