# PP Data run 9 Oct (BKK ~11:30)

## #45/#63 main disk (main 95% now, research-box 80% after others compressed)
- Verified: row counts (pools, swaps, transfers, bars, launches, kv) of all 56 main /opt/hist/rh day files are <= research-box copies (rb has 60 days to 10-08). Exception: main 2026-08-28.db is CORRUPT ("malformed", disk-full damage); rb copy is fine.
- Plan (script ready: /opt/crew/work/ppdata/slimrh.py): for days before 2026-09-25 replace each main day file with a slim copy holding only pools+kv (venuepick.first_seen_map reads pools of every day; aipick5 reads only the latest day). Keeps last 14 days full. Expected saving ~15 GB on main.
- NOT DONE: the auto-mode permission classifier blocked the replace/delete step as mass delete. Needs Ben/manager OK (or a permission rule) to run: `scp slimrh.py main:/tmp; ssh main 'nohup nice ionice -c3 python3 /tmp/slimrh.py &'` (counts in /tmp/cnt_main.json, /tmp/cnt_rb.json on main/rb). Log goes to /opt/hist/rh_slim.log.
- rhpull cron on main is PAUSED; it must stay paused (or exclude old days), else it re-copies the days.
- liqlock plan not started.

## #56 research-box ppcopy
- v4sender.db.tmp already gone. /opt/ppcopy/v4sender.db and /opt/sw/ppdata/v4sender.db are separate 13 GB files (different inodes). Plan: v4sender.py writes OUT straight to /opt/ppcopy/v4sender.db, old path becomes a symlink, delete the duplicate. Blocked by the same classifier on the delete. research-box is at 80% so not urgent.

## #58 gold_inc fails at tags step
- Cause: gold/creates.json (87,248,896 B) is TRUNCATED mid-record (last record blk ~22.6M, written 8 Oct 18:07 during disk-full); creates_head.json says next=83,476,700. tags.json itself parses fine (backup /tmp/tags.json.bak).
- Fix needs the creates list rebuilt for blocks >= ~22.6M (re-scan of PoolCreated, or from the chainlogs tape). Not done; hourly gold_inc will keep failing until then.

## #46 v3from.db: pdone 493 of 499, unit v3from inactive (after_chainlogs.sh still waiting for 84 tape files; chain/ has no .done files, b052.. still open). Not complete, not posted to analyst2.
## #16 #38 #51 #61: not touched this run.