Why are there hard timeouts on self hosted runners?

[Thomas Hepworth]

I use self hosted runners to do long FPGA builds from our companies bitbucket. 

There is a timeout of 12 hours for a pipeline and also a *shorter* timeout for queued jobs. I effectively cannot queue jobs since my 5 hour build takes longer than the queue timeout (~2.5 hours?).

I do not understand the justification for having timeouts when I host the runners on our own machines. It is our compute power to spend. Please could someone explain this to me, or how to bypass it? This is a major limitation which does not affect other platforms eg gitlab.

[Ben]

Hi Thomas,

Thank you for sharing your feedback on this.

We have had users raise this in the past, and we have an active feature request raised with our developers to be able to configure/increase the queued step timeout limit for runners. Please feel free to Watch this for future updates and Vote for it to improve its visibility with regard to customer demand:

•  https://jira.atlassian.com/browse/BCLOUD-23547

At this stage, the only workaround would be to create additional runners, this way an alternate runner from the pool could pick up the queueing job and prevent the build from failing.

Cheers!

- Ben (Bitbucket Cloud Support)