Skip to content

auto deploy: boot instead of switch#217

Draft
24apricots wants to merge 1 commit intomainfrom
24apricots-patch-2
Draft

auto deploy: boot instead of switch#217
24apricots wants to merge 1 commit intomainfrom
24apricots-patch-2

Conversation

@24apricots
Copy link
Copy Markdown
Member

switch causes problems sometimes, lets use boot by default instead. we can manually switch as needed until we implement an automatic method

todo:

  • switch for servers?
  • auto reboot for lab computers?

switch causes problems sometimes, lets use boot by default instead. we can manually switch as needed until we implement an automatic method

todo:
- switch for servers?
- auto reboot for lab computers?
@chamburr
Copy link
Copy Markdown
Member

what are the problems?

@BNH440
Copy link
Copy Markdown
Contributor

BNH440 commented Mar 30, 2026

what are the problems?

we just had an issue today where the tmpfs change we deployed with switch caused an sddm issue preventing login. this issue was fixed by a restart. the problem was caused by deploying on the live system. if we had deployed that update using boot, we should’ve never run into the issue.

so it makes sense to make the default boot so we don’t run into this kind of issue again, as we’re usually testing our changes after a restart, and can’t easily catch bugs introduced by applying the config live.

@chamburr
Copy link
Copy Markdown
Member

why would we want to disable switching permanently and make life difficult just because this astronomically unlikely event happened once

@24apricots
Copy link
Copy Markdown
Member Author

i think you misunderstand: switching is not permanently disabled. the goal is to make switch opt in so that system state isnt messed up in the middle of the day

@chamburr
Copy link
Copy Markdown
Member

i meant as in updates are never automatically activated. now if anyone makes a change and wants the change live they will either have to: (1) restart all computers, which is impossible even at 4am sometimes, or (2) manually run the switching command. (2) is what they will want to do 99.9% of the time so we are just adding more work with near-zero benefits — it's not going to happen again that someone makes an update to the filesystem, forgets to reboot computers, there's a process that broke because of it, and the process is critical enough that it will cause major disruptions

Copy link
Copy Markdown
Contributor

@BNH440 BNH440 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

one thing I think we forgot to consider, how are we going to update servers? as we'd like to always use switch for those as a restart each day probably isn't acceptable

can we do switch for servers and boot for desktops?

@24apricots 24apricots marked this pull request as draft March 31, 2026 23:09
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants