Bug #4809
closedkatello-jobs takes a really really long time to restart
Description
From MDP2:
- time service katello-jobs restart
Starting katello-jobs: [ OK ]
real 0m50.070s
From katello-nightly:
- time service katello-jobs restart
Starting katello-jobs: [ OK ]
real 2m18.895s
Updated by Mike McCune over 10 years ago
Also effects the 1st page render after a restart.
See strace logfile here:
http://mmccune.fedorapeople.org/scratch/strace-rails-1stpage.log.gz
Updated by Ivan Necas over 10 years ago
Getting rid of delayed jobs should get us back on the original re-start time. It takes twice as long now, as there are two workers started: dynflow and delayed jobs
Updated by Ivan Necas over 10 years ago
To give some backgound on the issue: there are two things:
1. delayed-jobs starting
2. dynflow execution starting
Both of them need to load the whole Rails environment to be able to do some work. We can move
away from dealyed-jobs quite soon, so I would not spend more time tuning the up
With dynflow executor, it would not help much not waiting for the environment to fully start,
as the Katello orchestration needs it to be running. So we would have faster start, but the
service would not be ready anyway (this is a trick that many init scripts do, but I consider
that cheating, causing more harm than benefit)
We could investigate why loading Rails environment takes so long and eliminate that as well.
I would help with unit testing as well.
Updated by Eric Helms over 10 years ago
- Related to Refactor #6297: Remove katello-jobs added
Updated by Eric Helms over 10 years ago
- Target version set to 55
- Translation missing: en.field_release set to 13