Bug 1653713
| Summary: | When the openvswitch vswitchd process is restarted in a compute node the vm conectivity is lost when using l2pop | ||||||
|---|---|---|---|---|---|---|---|
| Product: | Red Hat OpenStack | Reporter: | Candido Campos <ccamposr> | ||||
| Component: | openstack-neutron | Assignee: | Slawek Kaplonski <skaplons> | ||||
| Status: | CLOSED ERRATA | QA Contact: | Roee Agiman <ragiman> | ||||
| Severity: | high | Docs Contact: | |||||
| Priority: | low | ||||||
| Version: | 14.0 (Rocky) | CC: | amuller, bcafarel, bhaley, chrisw, skaplons, tfreger, tvignaud | ||||
| Target Milestone: | --- | Keywords: | Triaged, ZStream | ||||
| Target Release: | 14.0 (Rocky) | ||||||
| Hardware: | Unspecified | ||||||
| OS: | Linux | ||||||
| Whiteboard: | |||||||
| Fixed In Version: | openstack-neutron-13.0.3-0.20190119134915.886782c.el7ost | Doc Type: | If docs needed, set a value | ||||
| Doc Text: | Story Points: | --- | |||||
| Clone Of: | Environment: | ||||||
| Last Closed: | 2019-03-18 12:57:08 UTC | Type: | Bug | ||||
| Regression: | --- | Mount Type: | --- | ||||
| Documentation: | --- | CRM: | |||||
| Verified Versions: | Category: | --- | |||||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||
| Embargoed: | |||||||
| Attachments: |
|
||||||
|
Description
Candido Campos
2018-11-27 13:31:33 UTC
The problem seems be that the flows are not reloaded in the open vswitch when it is restarted *** Bug 1653319 has been marked as a duplicate of this bug. *** This looks like it could be https://bugs.launchpad.net/neutron/+bug/1775146 which was cherry-picked to stable/rocky in https://review.openstack.org/#/c/604494/ but not in our downstream branch. I will start a cherry-pick. @Brian: I'm not sure if that is the same issue like https://bugs.launchpad.net/neutron/+bug/1775146 - please note that here we are talking about restart of openvswitch and in U/S bug it was issue after restart of neutron-openvswitch-agent. fip conectivity is recovered restarting neutron metadata agent: [root@compute-2 heat-admin]# docker restart neutron_metadata_agent neutron_metadata_agent [root@compute-2 heat-admin]# Today I debug this issue with Candido on his deployment. It looks that after systemctl restart ovs-vswitchd we have 2 different issues: 1. problem with neutron-ovs-agent - it didn't configure properly open flow rules in br-tun thus private vxlan network didn't work - restart of neutron-ovs-agent fixed this. 2. problem with neutron-l3-agent - it didn't noticed that ovs-vswitchd process was restarted and didn't recreate qg-XXX port in qrouter- namespace, thus FIP connectivity didn't work. Restart of neutron-l3-agent container fixed this. Should we split this bug for 2 different ones then? Or should we track them together here? According to issue with vxlan networks in neutron-ovs-agent, I think that there is bug in L2pop mechanism driver: after ovs-vswitchd is restarted, it should reconfigure all fdb flows in the same way as it is done when neutron-ovs-agent is restarted but l2 mech_driver don't send all fdb flows to agent as this agent wasn't just restarted. I will need to check this issue on U/S code from master branch also to confirm but looking at the code now, I think that the bug is the same in U/S master and also in previous releases. So for bug with L3 agent I opened another bug: https://bugzilla.redhat.com/show_bug.cgi?id=1654371 Lets use this one to track issue with neutron-ovs-agent and L2pop mechanism driver. It looks like there are 2 issues here still: 1. With L2 population - I should have fix tomorrow 2. When openvswitch firewall driver is used, even if tunnels are set properly there is something missing in br-int because outgoing traffic from VM is blocked until neutron-ovs-agent will be restarted again. Created https://bugzilla.redhat.com/show_bug.cgi?id=1654489 for the other issue I found, let's use this for the issue described above. I found that there is indeed also issue with ovs-fw driver. I created separate BZ to track this one too: https://bugzilla.redhat.com/show_bug.cgi?id=1654643 Patch merged in U/S stable/rocky branch. Should be available in OSP-14 with first z-stream 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-2019:0585 |