mars
MARS-513
EVO-58 admin storage tile (Hostinger 25GB) — /administrar ambient panel
Blocked normal
cmcoder03-mars-ca
⛔ Waiting on bin-venus-cc to publish the storage-stats cron URL, then coder-mars-cc sets STORAGE_STATS_URL in Vercel production. Reclassified deferred -> blocked 2026-09-11 under pmmaster's fleet finding that setting a row to deferred CLEARS blockedReason, so this condition existed only in the description prose and was invisible to every status query. NOT VERIFIED either way: I could not read whether STORAGE_STATS_URL is already set, because the vercel CLI is absent on this host (the check returned a clean grep miss over a command that does not exist - absence of output was not evidence). Next hand: read the Vercel production env through the MCP before assuming the gate is still open; TERRA-36 sat deferred for 11 days behind gates that had both already been met.
Fleet EVO-58, Elazar-greenlit. Reads cron-published JSON {usedBytes,ceilingBytes,pct,ts,host} via STORAGE_STATS_URL env; shows GB-left+used-pct, warn tint >=75%, no per-app alert logic (cron owns the single 75% DM). Scaffold pushed 3cf9f01/v2.38.0, Class-S. Blocked only on bin-venus-cc posting the real cron URL -- coder wires STORAGE_STATS_URL Vercel env once relayed.
Activity
-
3cf9f01/v2.38.0 scaffold live, sin-datos fallback confirmed (unset/failed STORAGE_STATS_URL never breaks page). Awaiting real cron URL from bin-venus-cc to wire final env + close.
-
coder-mars-ca / coder
-
REASSIGNED to coder-mars-ca on coder-mars-cc's retirement 2026-08-24. STILL BLOCKED ON AN ELAZAR DECISION, not on work. bin-venus-cc measured the producer today: weekly timer on venus writes ~/.local/share/evolutiva/hostinger-storage.json, 130 bytes, exactly the five fields this consumer expects ({usedBytes, ceilingBytes, pct, ts, host}). NOTHING SERVES IT - no nginx location, no bind, local disk only. A Vercel function in sa-east-1 cannot reach venus local disk, so the URL this WI waits for will never exist. They relay an Elazar ruling (via coder-pluto-cc) that this is VENUS-ONLY: no tile, no transport, no publish maintainer; PLUTO-605 closed on it. bin-venus-cc EXPLICITLY declined to let a Mars WI close on a relayed second-hand ruling, and the incoming owner should hold that line too. I asked Elazar twice; unanswered at retirement. IF HE CONFIRMS: close as RULED OUT, never scope-obsolete - this is working, live, correctly-failing-open code whose INPUT was withdrawn by decision. Recorded that way or someone rebuilds the tile in six months. REVERSAL CONDITIONS bin-venus-cc recorded at the time, so a future reader gets more than the verdict: publish through PG where Vercel already reads (never a new public endpoint), name the registry-row maintainer FIRST, and set tile staleness at 8-10d because the cadence is 7d plus jitter - so absence means broken, never steady. AMBIENT NUMBER TO STOP REPEATING: the account is at 7.5%. The 72.6% figure attached to this work is the SHARED PHYSICAL VOLUME that disk_free_space reports on shared hosting, not our account, and it has been cited as urgency for this WI at least once.
-
coder-mars-cc / coder
-
DEFERRED by Elazar 2026-08-31 (asked directly, answered 'defer'). Not cancelled and not closed: the tile is live, correct, failing-open code whose INPUT was withdrawn - recording it any other way invites someone to rebuild it. Standing state: the venus weekly timer writes ~/.local/share/evolutiva/hostinger-storage.json (130 bytes, exactly the five fields this consumer expects) but NOTHING SERVES IT - no nginx location, no bind, local disk only - so a Vercel function in sa-east-1 can never reach it and STORAGE_STATS_URL cannot be set. bin-venus-cc declined to let this close on a relayed venus-only ruling and that line still holds; Elazar has now deferred rather than ruled, so do not close it on the relay either. REVERSAL CONDITIONS if it is ever revived (recorded by bin-venus-cc): publish through Postgres where Vercel already reads, never a new public endpoint; name the registry-row maintainer FIRST; set tile staleness at 8-10d because the cadence is 7d plus jitter, so absence means broken and never steady. AMBIENT NUMBER TO STOP REPEATING: the Hostinger account is at 7.5%. The 72.6% figure cited as urgency for this WI is the SHARED PHYSICAL VOLUME that disk_free_space reports on shared hosting, not our account.
-
Waiting on bin-venus-cc to publish the storage-stats cron URL, then coder-mars-cc sets STORAGE_STATS_URL in Vercel production. Reclassified deferred -> blocked 2026-09-11 under pmmaster's fleet finding that setting a row to deferred CLEARS blockedReason, so this condition existed only in the description prose and was invisible to every status query. NOT VERIFIED either way: I could not read whether STORAGE_STATS_URL is already set, because the vercel CLI is absent on this host (the check returned a clean grep miss over a command that does not exist - absence of output was not evidence). Next hand: read the Vercel production env through the MCP before assuming the gate is still open; TERRA-36 sat deferred for 11 days behind gates that had both already been met.
-
coder03-mars-ca
task
2026-07-19 by wi-cli-venus
4w ago