Re-evaluate max_restarts

[srid]

Should we make it a default in process-compose-flake? Or perhaps file an issue in process-compose itself asking what the recommend default should be, and why that default is not in process-compose itself.

Wasn't this used to be the case before? I vaguely remember process restarts never being infinite to begin with (I've seen processes enter into 'exited' state in process-compose window).

In any case, this change only affects services. Should we make it a default in process-compose-flake? Actually why not make it a default in process-compose itself?

Incidentally, there's no guarantee that a new service added to the repo will follow the same max_restarts convention. It is easy to miss that, so we should think about solving this more fundamentally.

Originally posted by @srid in #311 (review)

[shivaraj-bh]

How about having the common settings in

services-flake/nix/lib.nix

Lines 39 to 41 in 5e6aba2

	default = {
	namespace = lib.mkDefault config.namespace;
	};

Except for the exec.command most of the other things are common across services:

services-flake/nix/services/redis.nix

Lines 68 to 82 in 5e6aba2

	readiness_probe = {
	exec.command = "${config.package}/bin/redis-cli -p ${toString config.port} ping";
	initial_delay_seconds = 2;
	period_seconds = 10;
	timeout_seconds = 4;
	success_threshold = 1;
	failure_threshold = 5;
	};
	# https://github.com/F1bonacc1/process-compose#-auto-restart-if-not-healthy
	availability = {
	restart = "on_failure";
	max_restarts = 5;
	};
	};

[shivaraj-bh]

Or perhaps file an issue in process-compose itself asking what the recommend default should be

F1bonacc1/process-compose#242

[srid]

Ideal solution: upstream a sensible value for max_restarts.

If that is not practical, yes, we can make these common settings to be shared.