Nest with Fedora 2022 Infrastructure and Releng BARCAMP
schedule
2022-07-06 14UTC to 16:30UTC
14:00 to 14:15UTC - introductions and barcamp voting
14:16 to 15:00UTC - first block of discussions
15:00 to 15:15UTC - break
15:16 to 16:00UTC - second block of discussions
16:01 to 16:30UTC - revisting/roundup/smaller topics
discussion blocks - total 90min available, how shall we divide it. Please place an X where you prefer.
9 10min?
Votes:
6 15min?
Votes: XXX
3 30min?
Votes:
Discussion proposals:
Votes will be made until 14:15UTC. At that point the highest voted will be discussed first, then the next, etc.
Example talk - facilitator name / note taker name
description/more info
Votes: X
Notes:
5 year plans - Kevin Fenzi (nirik) / (mobrien)
I'd like to talk about where people see infrastructure going in the next 5 years? what will change? what can we do to be ready for it? Are there things we should focus on more? Is it all too volitle to tell? How about 3years? a year?
Votes: XXXXXX
Notes:
- We will follow some general trends such as moving to cloud, git ops etc. What can we do to prepare? Can we adopt things early like we did with puppet and Ansible?
- What do we think about writing documents as proposals for future work and use it as a guide/plan to follow as a roadmap for the future. A vision document rather than a set 5 year plan. This could help apprentices see what direction we are moving and what could help to get us there.
- Neal: 5 year plan is a bad idea! It generally doesn't work as is it difficult to manage specially with a 6 month release cycle.
- Examples: Document over wiki, migrate more to cloud, allow infrastructure to be easily updateable to allow for rapid upgrades to newer versions.
- Goals help us to be more proactive than reactive.
- Releng is also reactive and is closely tied with infra with large overlap
- If we wait for Fedora Council to lead on decisions we miss out on some thought leadership and remian reactive and proactive.
- There is a lack of people dedicated to releng which is a blocker towards proactive work
Questions:
- bookwar - Have you seen work by Alma Linux or Peridot on build systems? Alma uses Pulp as Artifact storage. Is something like this in scope for infra team?
Ideas:
How to handle Fedora Infra Tech Debt? (zlopez) / ?
As a Fedora Infra we accumulated plenty of Tech Debt during the years, some is more pressing, some not that much. I will share the current list of Tech debt we have and would like to discuss how should we address this or if you think anything is missing on the list.
Votes: XXXX
Notes: There’s a lot of tech debt. We have a spreadsheet.
python3 migration - pdc / FMN /
fedora messaging - FMN / badges
documentation
missing unit-tests
OIDC support
outdated systems
too many trjackers
not enough automation
unification of tools
monitoring
disaster and recovery plan
Major upgrades of deps
https://mobrien.fedorapeople.org/CPE Fedora_CentOS Tech Debt Matrix.ods
Ideas/Comments:
Questions:
How should Fedora Infra approach apps that we don't want to maintain? (zlopez) / ?
Three years ago we decided to get rid of some of the services we owned at that time. The reason for this was to better focus on the services that are critical for Fedora. Everything was explained in the blogpost, but we still have those apps around and community is using them, but there isn't anyone who actually maintains them. How we should approach this?
Votes: XXX
Notes:
- Updating README of apps that they are looking for maintainers.
- adopt a process for new apps like the package maint process (FTBFS/new maintainers, etc)
- standardize deployment so we can seperate knowledge of app from knowledge of infra
Questions:
Onboarding Improvement ideas (nirik) / (mobrien)
What can we do to improve our infra and releng onboarding? How can we be more welcoming? How can we get folks thru the 'a few changes' into a regular contributor? Can we do it without taking up tons of time from folks that are currently really busy?
Votes: XXXXXXXX
Notes:
-
Current onboarding: onboard during weekly meeting or mailing list. We add them to our fi-apprentice group which allows them read only ssh access to a number of servers which allows them to look around. We have docs which need work. Suggest some easy tickets to new users or a ticket which may be of interest to them. Not clear to jump from apprentice to higher access levels. We generally avoid direct mentoring mostly due to not enough mentors.
-
We should check on Openshfit metric access and if we can improve metrics and log accessibility
-
Infra as code is integrated strongly by wide use of Ansible
-
Provide some method to grant people titles within the community such as a "Fedora DevOps" or similar. Badges may help here
-
Permissions are hard to get and possibly to strict. Is it possible to grant access quicker? Possibly a more fine grained level.
-
How can community people help without access. Help with docs, attend meetings and ask questions. Tell us what knowledge you'd like to share
In chat - Maybe "proven packagers" can be approved also as "apprentice" by default?
Questions:
- (bookwar) does adoption of Openshift changes the onboarding practices? Can it help? More infra as a code approach? Easier access to Logs/metrics?
- Will more levels help? Not just apprenctice -> admin?
- Review-atons? Stream on Twitch?
- (zlopez) I remember talking about playground in past. Do we still want to do it?
Ways to improve communication (mobrien) / (zlopez)
Should we use mailing lists/ irc / discourse or something else. What are the best ways to communicate changes, discuss ideas or encourage new community engagement?
Votes: XXXX
Notes:
- We are not able to determine who is using what. I&R team doesn't use discourse that much.
- Mailing lists are useful, maybe mirroring them to discussion as read-only topics (announcements).
- Blogs are used for communication as well. Mirroring the blogs to other communication channels could reach more folks.
- Having central place that will be linked from other communication channels. Consumable in JSON format.
- Mailing lists vs. Discourse. Mails have customizable filters, Discourse is just giving it to you how they want.
- Preparing summary of what work we do. And sending it in consumable way.
- CPE Weekly is a lot of text - maybe making it easier to read. Same for quarterly reports.
Questions:
Questions and Answers (perhaps) (nirik) / ?
Questions from anyone on any infra topics.
Votes:
notes:
- communishift is coming back