CI congestion/starvation

[Heiko Becker]

Hello!

On Thursday, 12 February 2026 21:40:47 CET, Vlad Zahorodnii wrote:
> On 2/12/26 1:12 PM, Ivan Čukić wrote:
>> Can the bumps just contain a [ci skip] or similar? If it is a 
>> simple version
>> bump, does the CI need to run at all?

Just for the record, the preparations for the last Gear release happened at 
the end of January, not in the last 3 days. Framework was most likely 
before that, too. So at least in this particular case, it wouldn't have 
made a difference.

> That's the same thought that we had in Plasma. Every beta and 
> bugfix release used to clog CI so Bhushan did his amazing magic 
> to push Plasma version bumps with -o ci.skip. However, I believe 
> that other components still trigger CI. At least, we can feel 
> when apps bump their versions.

Personally, for Gear I felt that it's nice to have a CI run before the 
release, especially because there are modules, which don't have that many 
CI runs between releases. After all it's nice to catch issues before we 
ship them. Considering that Albert does his weekly failing CI runs, it 
seems less important though and I'd be willing to try using -o ci.skip.

However, and I don't have numbers about that, but I think we got a lot more 
CI jobs in recent times (let's say in the last year). So maybe we need to 
throw a bit more computing power at the problem, if we can afford it?

Regards,
Heiko