Files
archy/.claude/skills/podman/SKILL.md
T

90 lines
3.7 KiB
Markdown
Raw Normal View History

---
name: podman
description: Rootless Podman container management — diagnose, fix, and harden uptime. Use for container issues, port problems, UID mapping, health checks, or uptime hardening.
disable-model-invocation: true
allowed-tools: Bash, Read, Edit, Write, Glob, Grep
argument-hint: "[diagnose|fix|uptime] [container-name]"
---
# Podman — Container Management
Archipelago runs rootless Podman as `archipelago` user (UID 1000). All `podman` commands run without sudo. UID mapping: container UID N → host UID (100000 + N).
**SSH**: `ssh -i ~/.ssh/archipelago-deploy archipelago@192.168.1.228`
## Diagnose
```bash
# Container status
podman ps -a --format "table {{.Names}}\t{{.Status}}\t{{.Ports}}\t{{.Networks}}"
# Restart policies (must be "unless-stopped")
for c in $(podman ps -a --format "{{.Names}}"); do
echo -n "$c: "; podman inspect "$c" --format "{{.HostConfig.RestartPolicy.Name}}"
done
# Health checks
for c in $(podman ps --format "{{.Names}}"); do
health=$(podman inspect "$c" --format "{{.State.Health.Status}}" 2>/dev/null)
[ -n "$health" ] && [ "$health" != "<no value>" ] && echo "$c: $health"
done
# Resource usage + recent deaths
podman stats --no-stream --format "table {{.Name}}\t{{.CPUPerc}}\t{{.MemUsage}}"
podman events --filter event=died --since 24h 2>/dev/null | tail -10
# Rootless prerequisites
echo "XDG_RUNTIME_DIR=$XDG_RUNTIME_DIR" # must be /run/user/1000
grep archipelago /etc/subuid # must show archipelago:100000:65536
ls /var/lib/systemd/linger/ | grep archipelago # must exist
grep DEFAULT_FORWARD_POLICY /etc/default/ufw # must be ACCEPT
```
Cross-check 4 layers for port consistency: Backend config (package.rs) → Podman ports → Nginx proxy → Frontend appLauncher.ts. See `references/port-map.md`.
## Fix
**Restart policy missing**: `podman update --restart unless-stopped CONTAINER_NAME`
**UID mapping (permission denied)**: `sudo chown -R HOST_UID:HOST_UID /var/lib/archipelago/APP`. Formula: host_uid = 100000 + container_uid. See `references/uid-mapping.md`.
**Port conflict**: `ss -tlnp | grep :PORT` to find offender. Can't add ports to running container — must recreate.
**Network missing**: `podman network connect archy-net CONTAINER_NAME`
**UFW blocking LAN**: `sudo sed -i 's/DEFAULT_FORWARD_POLICY="DROP"/DEFAULT_FORWARD_POLICY="ACCEPT"/' /etc/default/ufw && sudo ufw reload`
**Stale processes**: `pgrep -c -f "podman ps"` — if >10, kill stuck processes.
See `references/common-failures.md` for the full error→cause→fix lookup table.
## Uptime Hardening
### Layer 1: Restart policies
```bash
for c in $(podman ps -a --format "{{.Names}}"); do
policy=$(podman inspect "$c" --format "{{.HostConfig.RestartPolicy.Name}}")
[ "$policy" = "no" ] || [ -z "$policy" ] && podman update --restart unless-stopped "$c"
done
```
### Layer 2: Watchdog timer
Create `/usr/local/bin/archipelago-container-watchdog.sh` that restarts stopped/unhealthy containers every 2 minutes via systemd timer. Script runs as archipelago user with `XDG_RUNTIME_DIR=/run/user/1000`.
### Layer 3: Ordered startup
Bitcoin stack has dependency chain: bitcoin-knots → electrumx + lnd → mempool + btcpay + fedimint → UI containers. Create `/usr/local/bin/archipelago-ordered-start.sh` with wait-for-container logic between tiers.
### Verification
```bash
sudo reboot # then SSH back after 3 min
podman ps --format "{{.Names}}" | sort # should match pre-reboot list
```
## Systemd Requirements
The archipelago.service needs these for rootless Podman:
- `ProtectHome=no` (podman stores in ~/.local/share/containers/)
- `PrivateTmp=no` (runtime in /tmp/podman-run-1000/)
- Do not set `RestrictNamespaces=` or `SystemCallFilter=`
- `Environment=XDG_RUNTIME_DIR=/run/user/1000`