This week's OpenDev PrePTG event resulted in a lot of successful discussion. We captured notes in the agenda Etherpad [0] alongside the agenda items and those interested should probably refer to that document after reading my summary. I'll try to keep this summary as brief as possible and will focus on actionable outcomes as a result.
At a high level we are comfortable with the code review practices we have been employing and will not be making any drastic changes. We did note that we should codify these practices a bit more than we have though. My first attempt at this would be:
* OpenDev code reviewers should be empowered to exercise judgement when applying single core reviewer approvals. It is OK to single core approve changes when reviewers are comfortable the change is low risk or low impact. * Holding off on approval despite a +2 is appropriate if the reviewer is unable to follow the change through merging and deployment. * It is generally a good idea to follow up on approved changes you have made to ensure they have deployed successfully. * Code pushed by another core reviewer is assumed to be implicitly approved by that approver unless they explicitly annotate the change with an indication it should not be merged (WIP/DNM/-2/etc).
We also discussed the possibility of adding new code reviewers through the TaCT SIG. Hopefully, a larger group of reviewers looking at OpenStack related configs will enable more time for OpenDev work from OpenDev core reviewers.
In the short term there are a number of tasks that we identified we could be working on:
* Clarkb will be reaching out to our cloud providers to reengage with them. We'll use this as an opportunity to determine if there might be new tools we can take advantage of (Kubernetes as a service for example), or if expected updates have been made (has OpenMetal's deployment tooling been updated). * We will proceed with upgrading MariaDB containers using the MARIADB_AUTO_UPGRADE env var flag. MariaDB seems confident that upgrades between the versions we run should be quick and painless. * The planned Prometheus service deployment can be picked up and completed. This would be a good task for a new contribute. * The Grafana container image can be unpinned. This is also a good task for a new contributor. * OpenDev git repo migration into the OpenDev Zuul tenant should continue. We think that the vast majority of our repos can either be moved or retired. Then we'll coordinate moves for system-config, project-config, and the DNS zone repos. * We can clean up Nodepool images and mirror content for OpenSUSE LEAP, Debian Buster, CentOS 7, and Ubuntu Xenial.
On a number of topics we realized that more investigation is required:
* The Podman, Docker, Docker Compose situation needs to be further investigated and tested. In particular OpenDev is interested in finding the least painful way to host images off of docker hub while still enjoying speculative container testing within Zuul jobs. Corvus is hoping to be able to look into this again. * Fungi will check if storyboard's container deployments can be resurrected in order to redeploy storyboard on the latest version of storyboard. Once that is decided we will be sending email about plans to sunset and retire the storyboard service. * We can use mirror web server logs to determine which mirror paths are most actively used. This information can then be used to decide on additional mirror content to prune.
We discussed a number of other topics without any solid conclusion. I want to call out two in particular that I would like the team to followup on, but please read the Etherpad nodes for complete coverage. First, is the question of how we want to handle email hosting for opendev.org. We can either host it ourselves or pay for hosted services which appear to be very reasonably priced. Second, is the management of some sort of backlog with indicators for new contributor and possible internship opportunities on the backlog items. Tonyb indicated a willingness to help mentor students involved in the Foundation's university partnerships through projects if we have good opportunities for them.
I am happy to answer any questions related to the event that you might have, or followup on agenda items I haven't already covered above. There was more discussion than I managed to cover in this one email, but I think it is long enough already. It might also be worthwhile to get other perspectives as mine is sure to have biases related to my work within OpenDev that are different than yours.
[0] https://etherpad.opendev.org/p/opendev-preptg-202402
Clark