uupd: fails due to network-online timing, Flatpak runtime missing, and registry pull timeouts

[hanthor]

DAKOTA 🦖  ·  issue pipeline
─────────────────────────────────────────────────
  ▶  filed      report received
  ·  approved   —
  ·  queued     —
  ·  claimed    —
  ·  done       —
─────────────────────────────────────────────────
  report:       missing     ·  confirms: 0
  area:         —         ·  priority: —
  next action:  same bug? ujust confirm 623

---Summary:
uupd (Universal Blue Update) fails to run successfully on host 'dilli'. Observed failures:

• Hardware check fails because network is not online: "Network, returned error: network not online".
• Flatpak updates error: "io.github.seadve.Kooha.Devel/x86_64/master requires the runtime org.gnome.Platform/x86_64/master which was not found".
• Composefs registry pulls time out: "pinging container registry 100.104.213.39:5001: ... i/o timeout".

Relevant log excerpts:

• May 31 09:44:53 dilli uupd[180209]: {"level":"ERROR","msg":"Hardware checks failed","error":"Network, returned error: network not online"}
• May 31 09:47:16 dilli uupd[182129]: {"level":"DEBUG","msg":"error: The application io.github.seadve.Kooha.Devel/x86_64/master requires the runtime org.gnome.Platform/x86_64/master which was not found"}
• May 28 16:02:45 dilli uupd[8204]: {"level":"ERROR","msg":"module_fail","output":{"Stdout":"error: Upgrading composefs: Checking if image 100.104.213.39:5001/dakota:chunked is pulled: Getting container info: Opening image docker://100.104.213.39:5001/dakota:chunked: failed to invoke method OpenImage: pinging container registry 100.104.213.39:5001: Get "http://100.104.213.39:5001/v2/\": dial tcp 100.104.213.39:5001: i/o timeout","Failure":true},"module":"System Update"}

Reproduction steps:

1. Trigger uupd (timer or service) before network-online.target is reached, or run on systems where network readiness typically lags.
2. Have Flatpak apps referencing EOL runtimes and composefs configured to pull from the registry above.

Suggested fix:
Add network-online.target dependency to uupd.service so updates wait for network to be ready. Create an override file:
/etc/systemd/system/uupd.service.d/override.conf
Contents:
[Unit]
Wants=network-online.target
After=network-online.target

Then:
systemctl daemon-reload
systemctl restart uupd.service

Also consider: enable systemd-networkd-wait-online.service or NetworkManager-wait-online.service as appropriate; make Flatpak module failures non-fatal or add retries for registry pulls; update/remove apps referencing EOL runtimes.

Notes: No reboot required; restart of the uupd service is sufficient.

Please triage and advise on whether the unit change is acceptable, or if uupd should be made tolerant of transient network/runtime errors.

[castrojo]

Can't reproduce