You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Establish new container lifecycle conventions across all HaLOS container apps to properly support multi-service compose files (sidekick containers).
Background
PR halos-org/halos-core-containers#5 adds a sidekick asset-server container to homarr. This revealed that the current container lifecycle conventions don't work well with multi-service compose files:
Previous convention (now being replaced):
restart: no - systemd manages the entire compose lifecycle
This breaks with multi-service compose files: if one container dies, systemd won't know (docker compose keeps running while one container is dead)
New convention:
Restart policy: restart: unless-stopped - Docker handles per-container restarts, systemd is fallback for compose process failures
Logging: logging: driver: journald - unified logging with per-container filtering via journalctl
Overview
Establish new container lifecycle conventions across all HaLOS container apps to properly support multi-service compose files (sidekick containers).
Background
PR halos-org/halos-core-containers#5 adds a sidekick
asset-servercontainer to homarr. This revealed that the current container lifecycle conventions don't work well with multi-service compose files:Previous convention (now being replaced):
restart: no- systemd manages the entire compose lifecycleNew convention:
restart: unless-stopped- Docker handles per-container restarts, systemd is fallback for compose process failureslogging: driver: journald- unified logging with per-container filtering via journalctlBenefits
journalctl -u <service>)journalctl CONTAINER_NAME=x)Implementation Plan
Related