I noticed several of the following errors in the first minute after ovnkube-node started. time="2019-12-03T01:05:41Z" level=error msg="Error while obtaining gateway router addresses for ip-10-0-168-35.us-east-2.compute.internal, stdout: \"\", stderr: \"ovn-nbctl: etor-GR_ip-10-0-168-35.us-east-2.compute.internal: port name not found\\n\", error: OVN command '/usr/bin/ovn-nbctl --private-key=/ovn-cert/tls.key --certificate=/ovn-cert/tls.crt --bootstrap-ca-cert=/ovn-ca/ca-bundle.crt --db=ssl:[2600:1f16:81b:4f00:c2f2:ccf4:70ff:2d8f]:9641,ssl:[2600:1f16:81b:4f01:2b16:3a57:788c:b3ed]:9641,ssl:[2600:1f16:81b:4f02:40ff:5c0b:5644:905c]:9641 --timeout=15 lsp-get-addresses etor-GR_ip-10-0-168-35.us-east-2.compute.internal' failed: exit status 1" time="2019-12-03T01:05:42Z" level=error msg="Error while obtaining gateway router addresses for ip-10-0-168-35.us-east-2.compute.internal, stdout: \"\", stderr: \"ovn-nbctl: etor-GR_ip-10-0-168-35.us-east-2.compute.internal: port name not found\\n\", error: OVN command '/usr/bin/ovn-nbctl --private-key=/ovn-cert/tls.key --certificate=/ovn-cert/tls.crt --bootstrap-ca-cert=/ovn-ca/ca-bundle.crt --db=ssl:[2600:1f16:81b:4f00:c2f2:ccf4:70ff:2d8f]:9641,ssl:[2600:1f16:81b:4f01:2b16:3a57:788c:b3ed]:9641,ssl:[2600:1f16:81b:4f02:40ff:5c0b:5644:905c]:9641 --timeout=15 lsp-get-addresses etor-GR_ip-10-0-168-35.us-east-2.compute.internal' failed: exit status 1" These errors happened several times at startup (once per second for 20-30 seconds). This comes from GatewayReady(), where ovnkube-node is waiting for the gateway router logical switch port to be created. ovnkube-master is creating it in GatewayInit() as it reconciles node resource state. Since this is a normal expected condition, it would be nice not to log these as errors in the log, making it look like a real problem.
Hi Just FYI concerning this, we had a discussion during out networking face2face regarding logging in OVN (see https://docs.google.com/document/d/11MvIjjmDy2sVpcej4b-tE2SMaSwhPtygmK_tO7-2StE/edit#heading=h.azeprfp4neg4) and we had agreed that we'd need to change all spurious error logs (which are not actually "errors", but more like "waiting for something to happen that has not happened yet"). This bug could either be fixed in US assigned to clean-up all logging in OVN, or the clean-up could be done during the resolution of this bug. /Alex
hi Moving this to ON-QA as a PR I made upstreams fixed those log messages: https://github.com/ovn-org/ovn-kubernetes/pull/1047/commits/7f7a02a93a0ec2e12aa1b5dcf48f5eaa96eb4d2a /Alex
Looks good to me on 4.5.0-0.nightly-2020-03-16-004817 Steps taken 1) Searched for error logs under all ovnkube-node pods like "oc logs ovnkube-node-28zvc -c ovnkube-node | grep -i error". Couldn't find any 2) restarted all ovnkube-node pods 3) Repeat step 1, seems good.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2020:2409