igneum-network/igneum
1
0
Fork 0
Code Issues Pull requests Releases Activity

release 0.3.22 section 13: the pull path and the 22:30 BST minute, build-1's three on 2720d8d2, the 0.3.23 Mac and HiveOS entries live at 22:31:50 BST, the 0.3.23 Windows take 1 findings, PC 1's helper death, the pool lane's 0.3.24 strings, the 1c420786 pairing ruling

Browse source 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This commit is contained in:
igneum-josh 2026-10-07 22:33:14 +01:00
parent f2b4dc0d3a
commit dae6b465e0
1 changed files with 10 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
docs/plans/release-0.3.22.md
View file

@ -128,3 +128,13 @@ Three commits on 96161037: bd710a36 the sub-version 3 re-pin (byte 7, igneum-pow
The 0.3.23 node's heights move Devnet 3's digest to ba75bf6f, and the one-box-at-a-time sweep to it split the network: dn3-g1 flipped at 21:18 BST and sat alone twenty minutes on its own fork (blocks 12,553 to 12,793, peers 0; the 83eb50cd nodes refuse ba75bf6f at the handshake; build-1's seed logged 27 digest-mismatch rejects by 21:37); the fleet stopped the pass at 21:31 after that one box (the Vast ssh proxies ssh1, 3, 4, 5, 6 refused the next five, so nothing else flipped) and reverted dn3-g1 to 34a2dbaa on its kept datadir, where it reorgs onto the network's heavier chain; no record or share of the network's sat on the solo blocks; the pool pair, the provers and the second nodes stayed on 83eb50cd. The shipper's part of the fault: the 0.3.23 Mac entry (both folders, 21:39:09 BST, DMG 4ada2807 on 7c7489ac) and the 0.3.23 HiveOS alias (0d716ed9, 21:37) were published ahead of the fleet's move, against rule 5's own word; both pulled back to 0.3.22 (live again at 21:41:08 and 21:44:48 BST; the Mac still runs 0.3.22; the 0.3.23 hive package held in scratch r0323/held-public). The move to 2720d8d2 is the 0.3.20 form (rule 5 extended): binaries pre-placed on every Devnet 3 node (the fleet's thirty, build-1's three, the pool pair's two), every node restarted inside one minute at a clock the fleet names ten minutes ahead once the proxies answer, the digest read back per node, the first lock on the new side as the line; the 0.3.23 Mac entry, the hive alias and the Windows entry publish at that minute or after. Pool split: the hour's 13,209 + 7,200 = 20,409 lands before the 0.3.24 publish minute, so the split goes into the 0.3.24 object at or after the v5 floor on IGNH/IGNW (pending main); the node lane holds it out of the v5 object commit. PC 1: the 5090 floor grid exited 0 at 21:41 BST but the Power Helper hung at its 1,000 MHz step and the 5090 mines at the 1,100 MHz lock until the hash lane's unelevated restore job runs; the grid gets the fix (stop at the first helper timeout, restart the task before the reset).
**Decimals (the founder, 21:40 BST): 18.** The testnet GO object re-cut goes on the 0.3.24 line as its second item on top of the v5 object commit (the 0.3.23 pin 2720d8d2 is frozen in the move): base_unit_decimals 18 with the UTXO/EVM scale fixed and tested, the 16-byte subsidy layout, the emission rescaled, the final 5 October message, the cache-rung field at 0, class v5 at 0 only if its Devnet 3 crossing reads clean, chain id 4462, the override file refused; the seeds re-armed on the dry run, nothing mining until the founder's go. **Pre-place for the one-minute move (the fleet, 21:5x BST):** the 2720d8d2 pair as .new on all 34 Devnet 3 nodes (20 rented, 14 standing second nodes), 2 of 34 placed while the Vast ssh proxies ssh1 to ssh9 refuse or time out since about 21:30; build-1's three pre-placed at 21:46 BST with the restart armed on the minute; dn3-g1 re-peered on 34a2dbaa at 21:43:32 BST (13 peers, three reorg lines, the 25 minutes of solo blocks orphaned). One miner per card restored by the fleet's v2 pass (p2-4090-3 first at 21:51:55 BST: the Devnet 3 miner alone at 55.96 MH/s, supervisors 4 to 1). **A pool-mode miner finding (0.3.24 item, the pool lane):** --exit-on-seed-change does not fire in pool mode; pool-b hashed the epoch-2 pack for 56 minutes after the 20:53 BST seed change (47,700 shares wrong_hash) until a re-export and restart by hand; the settled pool hour holds pool-b's shares to 20:53:09 only. PC 1: the Power Helper hang cleared by the hash lane's restore job at 21:45:58 BST (the 5090 back at 2,865 MHz); the 0.3.23 host slot open from 21:51 BST.
## 13. The one-minute move to 2720d8d2 and the 0.3.23 Mac entry (22:1x to 22:3x BST)
**The pull path (main's order, the fleet's build, live 21:59 BST):** every Devnet 3 box runs a puller that fetches a signed move file from build-1's /fleet/ each minute (the fleet ssh key, namespace igneum-fleet-move, verified against the key the box already trusts), downloads the 2720d8d2 pair as a tarball from /srv/artefacts, sha-checks it, posts FETCHED to the intake, and at the named minute restarts node and miner together from its running environment, posting APPLIED with the version, digest, peers and synced; the puller's install rides ssh once per box, after which a move never needs ssh; the dl host's moves/2720d8d2/ folder (both tokens, the hands pair f2cf6a87/fb147dd1, SHA256SUMS, 21:56 BST) stays the apps' path. **The minute: 22:30:00 BST**, published in the move file at 22:16 BST with 33 of 34 nodes fetched (dn3-g1 and g2 on the node-lane pair df6476ed/dfdc6883, every other node on the hands pair); dn3-j1 behind the dead proxy ssh6 falls off onto 83eb50cd alone and rejoins by the pull when its proxy answers (main's cut, named in the move file's notes); the new 3090 prover dn3-q05 applies on its first fetch. build-1's seed, node1-dn3 and observer restarted at 22:30:54 BST on the hands pair, each "igneumd/2.1.0-2720d8d2", digest ba75bf6f, the N15 latency-ladder line, 21 peers within the first minute. The fleet's per-node APPLIED lines and the first lock on the ba75bf6f side follow.
**0.3.23 Mac entry LIVE 22:31:50 BST (the copy at 22:30:28, the minute), both token folders, channel devnet-3:** Igneum-Miner-0.3.23-2720d8d2.dmg 4ada2807 (45,465,754 B; app 7c7489ac, the Mac node pair efff01cd/45c4c68f from 2720d8d2; packaged on the new token), interface 1.0.2 with ui_version stamped, the floor file kept, both manifests same fields, read back from both folders; the HiveOS alias on igneum-hive-0.3.23.tar.gz 0d716ed9 (the 2720d8d2 hive pair, both kit zips) at the same deploy; armed by r0323/publish-at-minute.sh on the fleet's minute (rule 5's digest-moving form). The Windows entry and the card wait on take 2.
**0.3.23 Windows take 1 FAIL (22:07 BST), install-close-23's own gap:** the installer sent the engine api/quit, the window host (alive since 21:20 BST) never stopped because stop-igneum.ps1 sits only in [UninstallRun] and CloseApplications cannot close it with the message box suppressed; the running 0.3.21 engine does not read the new install-running flag (expected) and relaunched 45 s after the quit from the host; every file stayed 0.3.21. Two more: the installer came out named Igneum-Miner-Setup-0.3.0.exe (BUILD-INSTALLER.bat passes no version; the job now passes -Version from version.h and asserts the output name; the batch fix is a packaging commit), and the stale flag is removed by a one-file job. The fix (the update-return lane): the .iss install path runs the stop step itself (the host by path after the quit, then the unlock wait), packaging only, the crate unchanged at 7c7489ac, so it rides 0.3.23; the kit re-cuts from it, take 2 in the same shape, the smoke read the gate (the known-failed shape: an install over a running 0.3.21 host, read on PC 2). The test the lane wrote asserted the stop script's text, not the install path's execution.
**PC 1 (22:2x BST):** the Power Helper task's process dies at once after each start since 22:08 BST (six "helper started" lines, no command processed, the task Running with no process; its last processed command 21:45, its idle exit at 22:05 clean; the app's tune idle), so the 5080 grids took unlocked rows only and the hash lane runs the lock-free jobs first (the 9070 XT G1 and ladder from 22:27 BST, then the family run and the v5 bench), the locked grids and the SM-sparse job returning when the helper answers; the helper's death is a 0.3.24 item for the update-return lane (the helper writes its reason before exiting, the engine shows "power control: helper not running (<reason>)"). The pool lane's 0.3.24 strings: fork pool-prepare-node 95ae3e50 on the v5 object tip (the pool-mode seed-change exit and the prepare deadline), daemon pool-finish-22 05f86a7f (the node connections reconnect by themselves after a node restart; the rejection counters), register MF-15 ff17216d; the split switch stays never. The 0.3.24 object pairs with the frozen class-v5 1c420786; the post-freeze fix 8ca66afa (AP-F4-1's weak-day redraw, the verified last resort; the pinned packs and fingerprint unchanged) is 0.3.25's.