Packit SRPM build takes too much time

[psss]

Several times I've observed that building the SRPM takes quite some time. I tried to do some basic time check on the teemtee/tmt#55 pull request. Here are the results:

15:20 ... commit pushed
15:21 ... travis started, packit SRPM build has just started...
15:23 ... travis tests completed, packit still building SRPM...
15:33 ... SRPM was built successfully (finally)

So it takes almost quarter of an hour to build the SRPM which is about 1 second on my laptop. Is there anything blocking the build? I think this delay should definitely be improved.

[TomasTomecek]

it seems like it's time to scale # of workers of packit-service up

[TomasTomecek]

We have now 3 workers so this should be resolved.

[psss]

Tested today on psss/edd#10. Looks better but still the difference is quite significant:

• local make srpm takes 1 second
• packit/srpm-build takes 5 minutes

Is there anything else making this simple initial step so slow?

[TomasTomecek]

This is why it's slow in the service (and sadly we can't change any of those, short-term):

• all the commands are running in a sandbox - the whole repo is being copied to the sandbox, this is usually a network operation which is brutally slow in the cluster
• it make take a while to get your job scheduled in packit-service
• in the new cluster, we no longer need to copy stuff - we'll be able to use network volumes and share the content like that

Long-term, we are planning on moving packit-service to a cluster with much better performance.

[psss]

I see, thanks for clarification. When do you plan to migrate to the new cluster?

[TomasTomecek]

as soon as possible, which will likely take weeks/months :/ since the new place is not even close to being ready

[psss]

Lately the response time seems to be better. Already moved to the new cluster?

[lachmanfrantisek]

Already moved to the new cluster?

No, unfortunately not yet. Maybe some small tweaks helped that.

[TomasTomecek]

Sadly, we're not migrating any time soon :/ we were promised a new place, but it didn't turn out in the end.

Maybe we could revisit this and try to do some optimizations on p-s' side.

[lachmanfrantisek]

I don't know why. But it behaves much better now.

Some optimization is always nice.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.
Thank you for your contributions.
We are doing this to be sure that the issue is still relevant. Anyone can comment to remove the stale state. (The issues marked with pinned, security, bug or EPIC label
are not considered stale.)