Re: [ovirt-devel] [VDSM] Flaky stompTests tests

Simone, The issue you are seeing is very interesting - a message we add to a deque disappear next time we check and according the log you provided there is no code accessing the deque. It happens only for one specific message. All the other messages work ok. Can you please gather logs so we could see what is really happening with it?

Thanks, Piotr 26 kwi 2016 10:18 "Simone Tiraboschi" <stirabos(a)redhat.com&gt; napisał(a): > 2016-04-25 20:15 GMT+02:00 Nir Soffer <nsoffer(a)redhat.com&gt;: > > Hi all, > > > > We see random failures in the stomp tests for long time. > > > > I suggest we mark these tests as broken for now. > > > > See examples at the end of the message. > > > > Nir > > > > ---- > > > > > 17:14:24 > ====================================================================== > > 17:14:24 ERROR: test_event(False) (stompTests.StompTests) > > 17:14:24 > ---------------------------------------------------------------------- > > 17:14:24 Traceback (most recent call last): > > 17:14:24 File > > > "/home/jenkins/workspace/vdsm_master_check-patch-fc23-x86_64/vdsm/tests/testlib.py", > > line 74, in wrapper > > 17:14:24 return f(self, *args) > > 17:14:24 File > > > "/home/jenkins/workspace/vdsm_master_check-patch-fc23-x86_64/vdsm/tests/stompTests.py", > > line 110, in test_event > > 17:14:24 client.callMethod("event", [], str(uuid4())) > > 17:14:24 File > > > "/home/jenkins/workspace/vdsm_master_check-patch-fc23-x86_64/vdsm/lib/yajsonrpc/__init__.py", > > line 340, in callMethod > > 17:14:24 raise JsonRpcNoResponseError(methodName) > > 17:14:24 JsonRpcNoResponseError: [-32605] No response for JSON-RPC > > event request. > > I'd suggest to investigate this a bit more since it can hide a serious > issue. > I'm moving hosted-engine-setup from XMLRPC to JsonRPC and I'm facing > exactly this kind of issue: it seams that some requests got lost and I > just receive a JsonRpcNoResponseError after a long time. > The real issue is that my request never reached VDSM getting lost > somehow in the queuing mechanism. > _______________________________________________ Devel mailing list Devel(a)ovirt.org http://lists.ovirt.org/mailman/listinfo/devel