Here is a list of things we may agree to merge.
Queue: use PostgreSQL notify for direct enqueue of jobs
Add facilities to parse the errors from the jobs so we can replace it by more contextual and helpful errors.
A logger which keeps in a buffer all the logs and flushes them when an error occurs in a synchronization, clears them if it succeeded
Job Channels: each job is owned by a channel and workers can be dedicated to one channel only