Big pile of fixes to adapt to current HEAD of core,scheduler,storage etc.
-
Indexers have been refactored so we now only one celery worker capable of doing all the indexers' work.
-
Fix the celery worker's hostname
-
Update config of the scheduler adapt to latest swh-scheduler's refactorings.
-
Add a prefix in celery workers' hostnames to prevent confusion eg. with pypi for which both the lister and the loader are named just 'pypi'.
-
Ensure pip, setuptools and wheel are up to date at image build time
-
Fix a typo in indexer's Dockerfile
-
Update config system for scheduler services
-
Workers' celery config is now managed as the [celery] section of the main config file
These 2 items below do not exactly fit in this pile of fixes, to be honnest... Blame lazy me!
-
Reduce the slicing for objstorage it's probably best in the context of adocker test env...
-
Small fix for the prometheus service
looks something has changed recently in the prom docker, not sure how to know what exactly, it's unclear to ne how the prom/prometheus published on docker's hub is actually built.
Migrated from D1052 (view on Phabricator)