The forking itself is/should be really quick what is slow is re-generating the ACLs on the backend and that's not an easyfix (and also highly depending on how many people are using the server at that time for example (ie how busy the workers are).

Can we get more workers? I appear to always be hitting busy times :)

We have this ticket https://pagure.io/fedora-infrastructure/issue/6441 to add monitoring to the workers so we can make an educated decision on whether we need more workers or not.

Note that there is always only one worker refreshing the ACLs on the backend to avoid multiple processes writing on each other. So more workers will make the fork come back quicker on screen but not the banner that ACLs are being refreshed go away more quickly.

ok - thank you. I am going to close this is as a duplicate.

Thanks :)

Also, we're working with upstream gitolite to try to speed things up (though in this case he already gave us a couple of proposals and we've been the ones lacking time to implement/test them :s)