Bug 2058379
| Summary: | [Docs] Write new "updating the ovn-controller" chapter | ||
|---|---|---|---|
| Product: | Red Hat OpenStack | Reporter: | kgilliga |
| Component: | documentation | Assignee: | kgilliga |
| Status: | CLOSED CURRENTRELEASE | QA Contact: | RHOS Documentation Team <rhos-docs> |
| Severity: | unspecified | Docs Contact: | |
| Priority: | unspecified | ||
| Version: | 16.2 (Train) | CC: | jamsmith, mciecier, sathlang |
| Target Milestone: | --- | Keywords: | Triaged |
| Target Release: | --- | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | If docs needed, set a value | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2022-03-24 12:35:04 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: | |||
| Bug Depends On: | 2050154 | ||
| Bug Blocks: | |||
|
Description
kgilliga
2022-02-24 18:50:21 UTC
@mciecier
I think that the KCS article you referenced provides more context as well. What do you think of the following update?
"Updating the ovn-controller container on all overcloud servers
Update the ovn-controller container to the latest RHOSP {osp_curr_ver} version. The update occurs on every overcloud server that runs the ovn-controller container.
[IMPORTANT]
You must update the ovn-controller containers on the RHOSP Compute nodes before you update the ovn-northd database on the Controller roles. However, until you update ovn-northd, ovn cannot create new ovs flows on the Compute nodes, which results in an outage on the control plane. The outage stops when the updated ovn-northd database becomes master on the cluster. An updated ovn-northd becomes master when you finish updating the second Controller node. Update all three Controller nodes to regain full network connectivity."
@sathlang please review the text as well.
Hi, I've updated the review, but will update as well that bz for consistency. > "Until you update ovn-northd, ovn cannot create new ovs flows on the Compute nodes. You can still create instances, but they won't have network connectivity. > The outage stops when the updated ovn-northd database becomes master on the cluster. An updated ovn-northd becomes master when you finish updating the second Controller node. It is recommended that you update > all three Controller nodes to regain full network connectivity." > ` so the kcs mentioned in the bz was for 16.1 and for when you use the external_ids:ovn-match-northd-version=true parameter, which was the original solution from OVN to support OSP style update, i.e. you apply that parameter on the OSP compute role and the ovn_controller will stop communicating with the ovn-northd on the OSP controller role until their version matches. That will prevent data plane cut (no ping loss) but you would have control plane loss (you cannot start new VMs) > Do we know sathlang if we are able to create vms with network connectivity after ```openstack overcloud external-update run --stack <stack_name> --tags ovn``` was run, but ```update run --limit controller`` is yet to be run? No, it's an needed improvement to the update testing. It should be there soon. Bottom line, you should remove from "However, until...." to final dot of that section. It shouldn't be the case and if it was it would be a critical bug and would need to be addressed. |