felhom.eu/documentation/audits/update-night-2026-09-21/failwalk.py

#!/usr/bin/env python3
"""failwalk.py <app> <sub> — what the brief says to do with a real edge that FAILED or HELD.

    "A failed or HELD real edge is the most valuable result of the night. Do not fix it, do not
     retry it twice. Evidence off, the hold sentence quoted in both languages, then walk the
     household's way out (restore per the sentence), record whether the data came back."

So this does exactly that and nothing else. It does NOT re-press the Update.

It re-seeds nothing: it uses the token file the failed run left behind where there is one, and
otherwise says plainly that the pre-update seed could not be carried across process boundaries and
re-seeds AFTER the restore instead — which proves the app works again, but not that THIS row
survived. The difference is stated rather than glossed.
"""
import json, os, re, sys, time

HERE = os.path.dirname(os.path.abspath(__file__))
sys.path.insert(0, HERE)
import walk as w            # noqa: E402
from fixtures import FIXTURES   # noqa: E402


def sentences(app):
    out = {}
    for lang, sfx in (("hu", ""), ("en", "?lang=en")):
        html = w.page(f"/apps/{app}{sfx}")
        open(os.path.join(HERE, "apps", app, f"held-page-{lang}.html"), "w").write(html)
        t = re.sub(r"<script.*?</script>", " ", html, flags=re.S)
        t = re.sub(r"\s+", " ", re.sub(r"<[^>]+>", " ", t))
        out[lang] = [s.strip() for s in re.split(r"(?<=[.!?]) ", t)
                     if any(k in s.lower() for k in
                            ("friss", "update", "vissza", "restore", "ment", "backup",
                             "masolat", "másolat", "copy", "meghajt", "hib", "error"))][:12]
    return out


def main():
    app, sub = sys.argv[1], sys.argv[2]
    d = os.path.join(HERE, "apps", app)
    os.makedirs(d, exist_ok=True)
    w.login()
    w.say(f"==== FAILWALK {app}: the household's way out of a held update")

    st = w.stack(app)
    held = {"state": st.get("state"), "updating": st.get("updating"),
            "update_phase": st.get("update_phase"),
            "update_phase_label": st.get("update_phase_label"),
            "update_error": st.get("update_error"), "hold_reason": st.get("hold_reason"),
            "observables": w.observables(app)}
    w.say(f"  [1] held state: phase={held['update_phase']} hold={held['hold_reason']!r}")
    w.say(f"      update_error = {held['update_error']!r}")
    w.say(f"      pinned    = {held['observables']['pinned_images']}")
    w.say(f"      installed = {held['observables']['installed_images']}")
    w.say(f"      compose   = {held['observables']['live_compose_image_lines']}")
    w.say(f"      inspect   = {held['observables']['docker_inspect']}")

    sent = sentences(app)
    w.say(f"  [2] hold sentence HU: {sent['hu'][:4]}")
    w.say(f"  [2] hold sentence EN: {sent['en'][:4]}")

    logs = w.app_logs(app, 400)
    open(f"{d}/held-app-logs.txt", "w").write(logs)
    mig = None
    for line in logs.splitlines():
        if re.search(r"Applying .*\.\.\. OK|migrat", line, re.I) and len(line) < 300:
            mig = line.strip()
    w.say(f"  [3] the last migration line the app printed, verbatim: {mig!r}")

    # THE WAY OUT — the button the sentence names
    way = w.restore(app)
    w.say(f"  [4] restore: {way.get('http')} in {way.get('seconds')}s; "
          f"state={way.get('state_after')} hold={way.get('hold_after')!r}")
    w.say(f"      pinned after  = {(way.get('observables_after') or {}).get('pinned_images')}")
    w.say(f"      inspect after = {(way.get('observables_after') or {}).get('docker_inspect')}")

    # did the data come back? asked of the APP
    fx = FIXTURES.get(app)
    after = {"route": None, "ok": None,
             "note": "the pre-update seed token did not survive the failed run's process, so this "
                     "re-seeds AFTER the restore: it proves the app WORKS again on the old version, "
                     "not that the specific pre-update row survived. Stated rather than glossed."}
    if fx:
        tok = fx.seed(w, sub, w.say)
        if tok is not None:
            after["route"] = "re-seeded after the restore"
            after["ok"] = fx.verify(w, sub, tok, w.say)
    w.say(f"  [5] the app answers its own front door again and holds data: {after['ok']}")

    sent_after = sentences(app)
    w.say(f"  [6] sentence after the restore HU: {sent_after['hu'][:3]}")

    rec = {"leg": "failwalk", "app": app, "held": held, "hold_sentences": sent,
           "last_migration_line": mig, "way_out": way, "data_after_restore": after,
           "sentences_after_restore": sent_after}
    json.dump(rec, open(f"{d}/failwalk.json", "w"), indent=2, ensure_ascii=False)
    open(f"{d}/failwalk-log.txt", "w").write("\n".join(w.LOG) + "\n")
    w.say(f"  written -> {d}/failwalk.json")


if __name__ == "__main__":
    main()