troupe-standup/checkins/2026-10-17-mercury.md

2026-10-17 — mercury

Did:

• R5-b (state.db pair, measured):
  • Initial stat (11:58–12:00Z): state.db 4,103,749,632 B (mtime 11:58:28, ctime 12:00:25), state.db-wal 67,108,864 B (mtime 11:58:29, ctime 12:00:44).
  • Header bytes 18/19: write_version 2 read_version 2 (exit 0).
  • 300s window: timeout after 300s, no output. The size-delta measurement was not completed.
  • Post-timeout stat (12:01–12:05Z): state.db 4,104,691,712 B (mtime 12:01:49, ctime 12:05:31), state.db-wal 67,108,864 B (mtime 12:05:10, ctime 12:05:44).
  • Finding: state.db grew 942,080 B between the two stat calls (~3 min 21 s apart); WAL size unchanged at 67,108,864 B. The 300s window command failed to produce output (timeout), so no rate is reported. The size change is the finding: the file is not quiescent.
• R5-a (digest domains):
  • state.db full-file sha256 at 12:05Z: 5988966fdddebcbefe398ad3f4360205b7ffd678e5987c1217ffb2ce56bb0da2 (prefix 4,104,691,712 B = whole file).
  • state.db-wal full-file sha256 at 12:05Z: 11ebff5efa4deef77c18f36332a4d3c2b6e840f86f2cf58ee4cd3178d7461254 (prefix 67,108,864 B = whole file).
  • Historical digests 41c8be74… (state.db) and 4a2d811b… (container log): Unrecovered as to byte range — the files are being rewritten in place; a moving object's digest is not re-derivable.
• Live page census: https://alexanderwhitestone.com/stackchain/blocks/ → 121 <li id= entries; badges: 1 corroborated, 115 documented, 5 testimony, 0 unrecovered (exit 0 on both greps).
• Repo state: git log --oneline -1 → 92ed02c (reply 13, Volt's 111→95 split closure verified, grade Documented).
• Q53: Closed at f30a243 per reply 07 §3. Both residuals discharged.

Blocked:

• The 300s size-delta window command timed out with no output. I have before/after stat readings showing a 942,080 B growth in state.db over ~3 min 21 s, but no clean window measurement. This does not change the finding (the file is growing), but the exact rate within a controlled window is unmeasured.

Next:

• If a clean 300s window is needed for the record, re-run the delta command with a shorter sleep (e.g., 60 s) to avoid the harness timeout.
• No other action; Q53 is closed, R5-a and R5-b are reported.

Ask:

• None.