Geo: Some Container repositories stuck in started state
Ticket https://gitlab.zendesk.com/agent/tickets/228449.
The customer sees that some containers were not synced for a while. We found that 154 registry records were in a started state, we've set them as failed with Geo::ContainerRepositoryRegistry.with_state(:started).update_all(state: :failed) => 154
and waiting for results but it's already clear that there is some bug.
Edited by Valery Sizov