This issue is caused by logrotate cron jobs that are started inside nailgun and keystone containers. Obviously, they can't reset nginx/postgres running in another containers. As a result, we got this problem.
The obvious way to fix this issue is to delete logrotate cron jobs from containers. It also possible to delete logrotate configuration from installed software, but it will take additional effort.
This issue is caused by logrotate cron jobs that are started inside nailgun and keystone containers. Obviously, they can't reset nginx/postgres running in another containers. As a result, we got this problem.
The obvious way to fix this issue is to delete logrotate cron jobs from containers. It also possible to delete logrotate configuration from installed software, but it will take additional effort.