Fedora Infrastructure Hackfest 2018 Recap

April 20, 2018  ·  fedora, travel

Last week was the 2018 Fedora Infrastructure Hackfest which took place in Fredericksburg, Virginia. I think everyone who attended found it to be a very productive week of hacking.

We started the hackfest off by brainstorming new documentation for packagers and those interested in becoming a packager. We decided to use asciidoc and put the documentation in a new git repo on Pagure. Right now it uses Pelican and Asciidoctor for generating html versions of the source files, but there was talk about eventually transitioning to Antora at some point. The point of the new documentation is that many of our workflows have changed with the move of packages/dist-git into Pagure. For example, it’s now possible to send pull-requests on packages and ask maintainers to merge them. So the documentation needed to be updated to reflect all of these changes, and it seemed like an ideal time to revamp and rework the documents.

We decommissioned darkserver and summershum, two apps that had been nigh-neglected for quite some time.

We worked on rawhide gating – that is, making it so that Rawhide updates must past through Bodhi. The big idea here is that we want to be able to let automated tests run over Rawhide updates before they are pushed out to Rawhide users, in order to make the Rawhide experience more stable. While Randy and Patrick did most of the Bodhi updates to make this happen, I began working on changes to the bodhi CLI. I am currently waiting until they have something for me to test my changes against.

On the same day that was going on, I also deployed the redirect to move the Fedora Jenkins setup into the CentOS infrastructure. It is now available here.

I also worked out a bug with the modernpaste paste-deactivation script I wrote for us to deactivate pastes when necessary. Recently, I redeployed modernpaste to Fedora 27 VMs, but the newer Python version caused an issue with the imports I used. Ultimately, I had to add a missing import and all was good again.

We talked about setting up Ansible AWX and came up with a plan and tried deploying it on a VM. I did most of the grunt work of setting up the VM, and Kevin and I worked together on the initial AWX attempts. Unfortunately, we ran into a fair number of what seem to be upstream bugs. In the interim, in an effort to at least get something up, we changed our deploy (e.g. by making it use a local database container instead of our external database VM), and eventually got it functional enough to play around with. Patrick started working on integrating OpenID Connect authentication into it, with the goal of users being able to log into our AWX with their FAS credentials. AWX seems like it would be nice in that we’d have more fine-grained control over which playbooks can be run by whom, however the pain of setting it up so far has been fairly off-putting. It will be interesting to play around with for a while longer and see what the team thinks.

My last day at the hackfest, Thursday, was discussion mainly about our OpenShift instances. We discussed moving more apps into OpenShift, plans to upgrade our instances to the latest releases, and so on. I signed on to move two more apps into OpenShift in the near future. Patrick gave us a demo of how it looks to use our current Ansible setup to deploy an OpenShift app. Over Thursday and Friday, he and Randy worked on moving Bodhi to OpenShift, but ultimately, due to several bugs in the Bodhi release, earlier this week, we reverted back to the old setup for the time being.

This is what I worked on. There were some other things going on as well. RHEL 7.5 is now synced in our repositories and ready for us to upgrade to it after freeze, along with OpenShift 3.9. Kevin worked on new priorities and templates for our Infrastructure ticket tracker.

Oh, and on Thursday night, I sang two karaoke songs, winning a $25 giftcard to the venue we were at for dinner/drinks that night. I ended up giving it to one of the other singers, since I don’t live in Virginia, and will likely never be back at the venue.