https://jira.opendaylight.org This file is an XML representation of an issue en-us 8.20.10 820010 22-06-2022 https://jira.opendaylight.org/browse/CONTROLLER-1841 controller <p>the robot test case to find the new leader normally first polls on /restconf/modules<br/> before going to jolokia to check the car shard for leadership. in some failure cases<br/> we notice that the polling goes for 35s on /restconf/modules (polling 5 times with<br/> 5s timeouts) and fails. We want to skip that check and just go straight to /jolokia<br/> just to collect that data point.</p> <p>/restconf/modules should still respond, imho, so we would want to figure that out<br/> eventually as well</p> CONTROLLER-1841

skip checking /restconf/modules

Sub-task CONTROLLER-1838 Medium Verified Done Jamo Luhrsen Jamo Luhrsen Fri, 22 Jun 2018 17:17:22 +0000 Thu, 5 Jul 2018 20:14:27 +0000 Thu, 5 Jul 2018 20:14:22 +0000 clustering 0 4 <p>When I remove the initial /restconf/modules check from the logic that figures out who the new leader and followers<br/> are after isolation, those steps are passing. However, there are other REST interactions in these tests that will fail<br/> for a similar reason.</p> <p>I think at the end of all of this, we really need to understand and fix what is causing /restconf to be unresponsive.<br/> That is too basic of a problem to ignore, and is just getting in the way of other debugging we might need to do.</p> <p>For example, after skipping the /restconf call when figuring out the new leader, the test just fails when trying to use<br/> restconf to add cars, and then again when trying to read from restconf to count how many cars are added.</p> <p>If we are lucky, and this tell-based protocol is doing what we expect, then maybe this unresponsive /restconf/<br/> issue is our only problem in this job to worry about.</p> <p><a href="https://logs.opendaylight.org/releng/vex-yul-odl-jenkins-1/builder-copy-sandbox-logs/138/jamo-controller-csit-3node-rest-clust-cars-perf-only-fluorine/3/" class="external-link" target="_blank" rel="nofollow noopener">example job</a></p> <p>we did try this, and it did help somewhat as we didn't get quicker failures, but when the restconf was getting locked up (401 or 500) the csit will eventually hit the problems in some fashion<br/> (all ODL interaction is via REST)</p> Development Rank 0|i03fvz: