Old application is stopped while new application is pulling

[hrfuller]

When we update an application spec with a new image, the docker container associated with the old image is stopped while the new one is pulled, leading to downtime. We discussed this on discord, but I'm making a ticket where I can post logs, and other diagnostic info.

I have attached the synpse-agent debug log here. Critically you can see that the running container is stopped after the update.

18 14:34:54 <redacted-host-name> synpse-agent[828]: 2023-10-18 14:34:54.404        DEBUG        info/reporter.go:121        workload status changes detected, synchronizing        {"version": 971, "resource": "Application"}
160 <18 14:34:54 <redacted-host-name> synpse-agent[828]: 2023-10-18 14:34:54.404        INFO        info/reporter.go:173        app reconciled        {"name": "<redacted-application-name"}
161 <18 14:34:54 <redacted-host-name> synpse-agent[828]: 2023-10-18 14:34:54.589        INFO        docker        docker/handle.go:122        stopped container        {"container_id": "4daaeded461993b81fd47f4226991115e5abe908d359f43980f1b4753c5ab693"}

synpse_update_logs_abridged.txt