Support for :max-retries in restart

[d0b3rm4n]

The docker-compose allows to set a a max-retries in the the restart configuration.

https://docs.docker.com/reference/compose-file/services/#restart

on-failure[:max-retries]: The policy restarts the container if the exit code indicates an error. Optionally, limit the number of restart retries the Docker daemon attempts.

Having a compose file like:

services:
  db:
    image: docker.io/library/postgres:17
    restart: on-failure:5

causes an error like this:

$ podlet compose container-compose.yml
Error: 
   0: error converting compose file
   1: error reading compose file
   2: File `container-compose.yml` is not a valid compose file
   3: unknown variant `on-failure:5`, expected one of `no`, `always`, `on-failure`, `unless-stopped`

Location:
   ~/.cargo/registry/src/index.crates.io-1949cf8c6b5b557f/podlet-0.3.1/src/cli/compose.rs:219

Would be nice to get support also in potlet, thanks.

[TheRealBecks]

I tried to find the Systemd and Quadlet settings that map this behavior:

• The restart: on-failure has to be set with a Systemd service setting: Restart=on-failure
• Additionally there are the StartLimitIntervalSec=interval and StartLimitBurst=burst settings
• In addition to the linked compose spec theres also the [Docker behavior](https://docs.docker.com/reference/cli/docker/container/run/#restart) that explains how the restart` works:

An increasing delay (double the previous delay, starting at 100 milliseconds) is added before each restart to prevent flooding the server. This means the daemon waits for 100 ms, then 200 ms, 400, 800, 1600, and so on until either the on-failure limit, the maximum delay of 1 minute is hit, or when you docker stop or docker rm -f the container.

If a container is successfully restarted (the container is started and runs for at least 10 seconds), the delay is reset to its default value of 100 ms.

And additionally:

If the ... container exits with a non-zero exit status more than 10 times in a row, Docker stops trying to restart the container. Providing a maximum restart limit is only valid for the on-failure policy.

@d0b3rm4n @k9withabone I don't think that this compose spec can be directly mapped to Systemd service and unit files. I also don't think that it makes sense to use StartLimitIntervalSec=interval and StartLimitBurst=burst as they only provide one interval that can't be doubled per failure. Also a "max seconds" value is missing.

@k9withabone That being said I would recommend to handle the :5 setting to prevent Podlet from failing. Maybe it's best to simple ignore it as it can't be mapped to Systemd?

[d0b3rm4n]

I tried to find the Systemd and Quadlet settings that map this behavior:

@d0b3rm4n @k9withabone I don't think that this compose spec can be directly mapped to Systemd service and unit files. I also don't think that it makes sense to use StartLimitIntervalSec=interval and StartLimitBurst=burst as they only provide one interval that can't be doubled per failure. Also a "max seconds" value is missing.

When I wrote "support" I left it open on purpose how that should look like. I myself was also investigating a little bit further after I filed the issue and could also not come up with a way how to solve that in Systemd. There is the --restart=policy option in podman create and podman run command, which allows a policy like on-failure[:max_retries]. But the man page says clearly not to use it if ran with Systemd:

When running containers in systemd services, use the restart functionality provided by systemd. In other words, do not use this option in a container unit, instead set the Restart= systemd directive in the [Service] section.

@k9withabone That being said I would recommend to handle the :5 setting to prevent Podlet from failing. Maybe it's best to simple ignore it as it can't be mapped to Systemd?

I agree best would be not to fail on the the :[max-retries] part, and ignore it. Nice to have would be some warning that this could not be mapped.

[k9withabone]

Huh, it seems that the Docker docs are out ahead of the official Compose specification (which is what the compose_spec library is based on) on this. I'll open a PR in compose-spec/compose-spec and an issue in k9withabone/compose_spec_rs to track it and add support.

[k9withabone]

That being said I would recommend to handle the :5 setting to prevent Podlet from failing. Maybe it's best to simple ignore it as it can't be mapped to Systemd?

I agree that ignoring it is probably the best solution.

[k9withabone]

Created compose-spec/compose-spec#635.