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

release-0.3.11 plan: the push and both CI runs, step 1 (the hand nodes and the seed at 4d8f8bb6, the manifest with the override carried over, the Mac, PC 2 and its epoch-boundary recovery, the stall and the Mac's resume, the laptop), the next-cut list

Browse source 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This commit is contained in:
igneum-labs 2026-10-05 23:56:46 +00:00
parent d85f4dec2b
commit f7613e5f96
1 changed files with 39 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

39
docs/plans/release-0.3.11.md
View file

@ -134,6 +134,44 @@ Three digests on the 0.3.11 binary, so two sweeps: step 1 moves every node to th
nine-field object (0139ab9d...). A node on either side of a sweep is refused by the other side (the handshake), so each sweep is one nine-field object (0139ab9d...). A node on either side of a sweep is refused by the other side (the handshake), so each sweep is one
window, as the fee switch's 8 min 47 s was. window, as the fee switch's 8 min 47 s was.
## 6. The push and CI
`git push -u origin release-0.3.11` at 3b0262f (23:18:29Z, the credential helper; the pre-push hook's site flip restored), `gh workflow run
windows.yml --ref release-0.3.11` -> run 37387737179, acquired at once (GitHub operational again), green 23:23:11Z (the parse job 23:18:39 to
23:19:25Z; engine, window host, payload, installer, smoke run 23:19:31 to 23:23:11Z, the G13 step against the 89dfcb95 inputs). The main
`ci.yml` run on 3b0262f (37387751432) failed in one step, the igneum-census release build (the class v3 fields missing from the census's own
initialisers, `fetch`'s new Layout argument, no Scratch/Hot match arms; pow tests, simulators and site jobs passed); the coordinator fixed it
in this worktree as 2a62735 (`igneum-census/src/main.rs` only; `git diff --stat 3b0262f 2a62735 -- app packaging igneum-pow proto-cuda
proto-opencl proto-metal` is empty, so the Windows artefacts of 37387737179 stand) and its run 37388453875 is green (pow tests and census
build, simulators, site). The 0.3.11 CI verdict is therefore run 37388453875 on 2a62735; the Windows build is run 37387737179 on 3b0262f,
the same app sources; the merge to master goes from 2a62735.
| File | sha256 | Size |
|---|---|---|
| Igneum-Miner-0.3.11.dmg | b7e81d4f6f3af9f9179e29faa72c844b795df757dfb2e4cd1d56cd454e78e1e7 | 41,592,041 |
| Igneum-Miner-Setup-0.3.11.exe | 84a21443f78598597f33cef307fa162e53c680dd90d29f02ceaf79ac5e231ee4 | 50,065,009 |
| igneum-windows-app.zip | ed0cf2a75a1de8897de33ddcdcdb4ea844eca6b38627d6b91c562700c9fbe5eb | 72,070,333 |
| igneum-hive-0.3.11.tar.gz (dl/public) | c606a17043c013c411086b4abd0f38a227aa8a7db9d5934157760a3da5a5edc9 | 24,496,653 |
## 7. The rollout, step 1 (the binary sweep to 4d8f8bb6...)
Baseline 23:19:40Z: tip DAA 139,642; the observer and node 1 on 21d4c73c at 1f4b4425...; the Mac app 0.3.10 (attached to node 1); PC 2
0.3.10 at 120.8 MH/s; PC 37ba0461 back on 0.3.10 at 2.3 MH/s; PC 1 down since 22:31:06Z (section 3a); Sam's Mac quit since 20:47Z.
| Step | Time | Result |
|---|---|---|
| 1a the observer | 23:23:54Z (pid 61754) | `igneumd/2.1.0-89dfcb95`, the four-field file, digest 4d8f8bb668828a3dcf7b783b995f3d3ebfde32a092dd1dbd5bf4373c5c65a62c |
| 1a node 1 | 23:24:06Z (pid 61864, caffeinate 61866) | the same |
| 1a the seed | 23:24:25Z (MainPID 125853) | `igneumd/2.1.0-89dfcb95`, the same lines; the a24ab01a... no: the 21d4c73c binary kept as `igneumd.prev-035` (the script's name) |
| 1b the ship (`--from ci`) | 23:24:44Z | preflight ok (tree 2a62735 clean, 0.3.11 in all 6, fork 89dfcb95, gh igneum-labs, live inputs 89dfcb95 built 23:17:51Z); ci, fetch, dmg, copy already; `consensus.override: carried over from the current manifest` (the four-field object), `activation_height` 135200, `deadline_note` "finality v3"; manifest 0.3.11 mac+windows signed (key 8f186e37...) and in `dl/public/`; one deploy; verify: the token folder's manifest 0.3.11, signature ok, mac b7e81d4f..., windows 84a21443...; the public folder's `igneum-downloads.json` not yet the local bytes at the edge (the 0.3.10 class), resumed from `verify` for the console item |
| 1c update-now, the Mac and the laptop | 23:28:06Z (`update-now-0311-d937c69d-37ba0461`, apps woken) | the Mac: the job ran 23:28:48Z, `Igneum Miner 0.3.11 is available: downloading (41 MB)` 23:28:49Z, engine restart 23:29:05Z (run `mac-d937c69d-20261005-232905`), 59 s after the job; `[ok] updated to Igneum Miner 0.3.11 from 0.3.10`; `a node already answers on 127.0.0.1:26610; using it`: the app attaches to node 1 (89dfcb95, 4d8f8bb6...), so its card follows node 1; `cards: Apple M5 Max [apple, off]`: its Metal miner was off before and after, so the prepared-pack check has no subject on the Mac tonight. The laptop (PC 37ba0461): (pending) |
| the console | 23:30Z | item #366 "Igneum Miner 0.3.11 shipped (mac+windows)" (`--from console` after the public index settled at the edge; the ship's own verify had refused it as 0.3.10's did) |
| the old side during the window | from 23:24Z | PC 2 and the laptop at 0 peers (refused by the new side) until each updates. The new side (node 1, the observer, the seed, the Mac app attached to node 1 with its miner off) has NO miner on it, so its chain STALLED at DAA 139,751 from about 23:29Z (the observer's `/api/stats` at 23:31:42Z: 139,751, age 1.8 s) until a miner joins it; the old side's fork grows only while a miner mines there, and PC 2's 0.0 MH/s from 23:29Z is the prover-floor sweep stopping its miners for its run (`floor-sweep-2`, started 23:21:17Z), not the refusal. Put to the Counter ASIC coordinator at 23:31Z with the numbers; its call: "PC 2 clear" the minute the sweep closes (about 23:36Z) and at 23:45Z at the latest, the restore job after the update, because an app restart under the sweep kills its job tree and its restore never runs; the laptop's 2 MH/s restarts the new side's chain when its install ends. The lesson for the next cut's plan: step 1a (the hand nodes and the seed first) moves the hub to a side with no hash until the first miner updates; the first update-now should go to a miner within the same minute |
| the Mac's miner (C42) | 23:39:10Z | the new side had NO miner: the Mac app's mining was a persisted `settings.paused` (cleared only by Resume; its STATUS lines read "0.00 MH/s, paused" since the update), node 1 runs no miner, the fleet's hash was on the refused old side. `POST <app.url>/api/resume` (the token path) answered ok at 23:39:10Z; the Metal worker compiled the program inline at 23:39:11Z ("not prepared": the prepared-pack path did not engage on this start, a G4b finding, section 11), raced 14 variants, 3.15 MH/s wall at 23:39:48Z; the new side's first blocks accepted 23:39:49, 23:39:51, 23:40:00Z. The stall: about 23:29Z to 23:39:49Z, 11 minutes of stopped DAA clock on the side every node ends up on. New check for step 1 of every digest-flipping cut: the new side has at least one miner before the first update-now |
| 1c update-now, PC 2 | 23:39:44Z (`update-now-0311-1ccfe586`, on the Counter ASIC coordinator's "PC 2 clear": `floor-sweep-2`'s first point had hung 18 min, nothing in flight to protect; its restore-and-diagnose job is the first PC 2 job after the update). PC 2 fetched the woken file at 23:40:13Z and logged `1 new for this machine, 1 queued`: update-now is itself a job and the app runs jobs one after another, so it waits behind the hung `floor-sweep-2` (started 23:21:17Z, cap 30 min, freed about 23:51:17Z); PC 2 reads `0.00 MH/s, waiting | node 139753 blocks, 0 peers, syncing` every 30 s meanwhile (refused by the new side, its miners stopped by the sweep). The class, for the next cut's plan: an update-now cannot pre-empt a running job; a hung job's cap sets the update's time | the queued job ran at 23:51:19Z, two seconds after the sweep's cap; installer verified 23:51:22Z; `quit: stopping the miners, then the node` 23:51:24Z; engine restart 23:51:30Z (run `win-1ccfe586-20261005-235130`), 11 min 46 s after the job and 3 min 35 s after PC 2 had received it; `igneumd started` 23:51:32Z; the prover's host and the pinned ids as before; `miner nvidia-1ccfe586-1 started` 23:51:43Z. Check 1 PASS at the first start: `epoch seed f4d9d3d8... (daa 140361): CPU program and cache ready in 166 ms`, `worker: ready cuda NVIDIA_GeForce_RTX_5090 ... first pack ... self-test PASS` at 23:52:21Z, no refusal. Then the hourly epoch boundary fell at DAA 140,401 (`SEED CHANGE ... f4d9d3d8... -> cb5b51cc...`, 23:52:21Z, 40 s after the first pack): the worker refused the new epoch's jobs (`epoch seed mismatch: this worker holds epoch f4d9d3d8..., the job is for epoch cb5b51cc...`, 80 lines over 80 s) while the app's attempt-aware path prepared the new epoch (`program pack checked for epoch seed cb5b51cc... day 20732: attempt 0`, the line the pack-loop fix added; `PREPARE sent ... class v2 (3559 DAA blocks before the boundary at 144000)`), the worker switched at 23:53:42Z, STATUS 113.4 MH/s wall at 23:54:14Z, the card 112.4 MH/s at 23:56:04Z. The Mac's worker crossed the same boundary in 241 ms (`prepared cb5b51cc... class v2`). PC 2's node joined the new side at once (its DAA tracks the observer's, 1 peer: node 1 at 192.168.68.64). The sweep's restore-and-diagnose job is the Counter ASIC coordinator's, after this |
| the laptop (PC 37ba0461) | silent since 23:28:28Z | its last upload, "2.25 MH/s, mining \| node 139757 blocks, 0 peers" at 23:28:28Z, is 22 s after `update-now-0311-d937c69d-37ba0461` was published; no job line reached the intake before the silence. Its 0.3.10 install kept it silent 55 minutes (21:41 to 22:36Z), so this is its install in progress until shown otherwise; publish 2 does not wait on it (a 2 MH/s machine whose 0.3.11 reads the nine fields when it returns; on 0.3.10 its node would die on the file until the forced apply, the C39 case for one machine) |
| PC 1 | unreachable tonight (section 3a); refused by every peer on 1f4b4425 until its morning relaunch takes 0.3.11 through the manifest | |
## 10. The next cut ## 10. The next cut
| Branch | What | Why not 0.3.11 | | Branch | What | Why not 0.3.11 |
@ -141,6 +179,7 @@ window, as the fee switch's 8 min 47 s was.
| `ember-tune` b671c8b (and the tune behind it) | the C35 fix: `Cmd::Quit(&'static str)` so every "quit:" line names its sender (the window host's stdin, the host gone, `POST /api/quit`, the sweep's end), `elevation_allowed()` = Power control alone (the unattended sweep on PC 1 raised one UAC prompt at 22:30Z under the old rule), no power cap at start under `--sweep`; the quit-source hunk is separable (main.rs 2 lines, server.rs 1 line, engine.rs the Quit arm, `elevation_allowed` and its test) | arrived after the tree closed at 23bc2b2 (the app, the DMG and PC 1's job carry it); not among the branches named for this cut | | `ember-tune` b671c8b (and the tune behind it) | the C35 fix: `Cmd::Quit(&'static str)` so every "quit:" line names its sender (the window host's stdin, the host gone, `POST /api/quit`, the sweep's end), `elevation_allowed()` = Power control alone (the unattended sweep on PC 1 raised one UAC prompt at 22:30Z under the old rule), no power cap at start under `--sweep`; the quit-source hunk is separable (main.rs 2 lines, server.rs 1 line, engine.rs the Quit arm, `elevation_allowed` and its test) | arrived after the tree closed at 23bc2b2 (the app, the DMG and PC 1's job carry it); not among the branches named for this cut |
| `fud-close` (the ledger closer's main branch, a22ba27a60a6f1c64; its ready tip was due about 23:05Z) | 45 public-text fixes on the site and litepaper, spec 8.3 and 8.8, two CI checks, relay fixes; touches `packfile.h` and `host.c`, so taking it means the two Windows workers, the Mac worker and the DMG rebuilt from the merged tip (G5) | offered by the Counter ASIC coordinator at 22:5xZ after the tree closed; not among the branches named for this cut; the fork-side `ledger-fixes` is not in 0.3.11 either | | `fud-close` (the ledger closer's main branch, a22ba27a60a6f1c64; its ready tip was due about 23:05Z) | 45 public-text fixes on the site and litepaper, spec 8.3 and 8.8, two CI checks, relay fixes; touches `packfile.h` and `host.c`, so taking it means the two Windows workers, the Mac worker and the DMG rebuilt from the merged tip (G5) | offered by the Counter ASIC coordinator at 22:5xZ after the tree closed; not among the branches named for this cut; the fork-side `ledger-fixes` is not in 0.3.11 either |
| `explorer` d7e797c (and 3e01212) | `/api/stats` gains `proving`; `tools/ci/public-api-check.mjs` then FAILS when the live API lacks it, and ci.yml runs that check against the live site on every master push, which reads the OLD API until Vercel redeploys after the push (the reviewer's C36) | not in 23bc2b2 (only on the explorer branch); its merge needs the check to retry for a few minutes or to require `proving` only when `observer_updated_at` is newer than the commit | | `explorer` d7e797c (and 3e01212) | `/api/stats` gains `proving`; `tools/ci/public-api-check.mjs` then FAILS when the live API lacks it, and ci.yml runs that check against the live site on every master push, which reads the OLD API until Vercel redeploys after the push (the reviewer's C36) | not in 23bc2b2 (only on the explorer branch); its merge needs the check to retry for a few minutes or to require `proving` only when `observer_updated_at` is newer than the commit |
| the app's job queue | an update-now job pre-empts a running job instead of queuing behind it, or the queue reports its wait in the STATUS line (tonight PC 2's update waited 11 minutes behind a hung sweep's cap, section 7; the Counter ASIC coordinator's ask) | app change |
| `proving-v1` c36dfea | docs only, after the code tip 22c2363 | its agent's choice: docs follow | | `proving-v1` c36dfea | docs only, after the code tip 22c2363 | its agent's choice: docs follow |
| the 0.3.10 list (`release-0.3.10.md` section 11): fork `pack-loop` 05ef0fa3, `job-console`, the rest of `opencl-rdna4`, `opencl-rdna4-telemetry` | | unchanged | | the 0.3.10 list (`release-0.3.10.md` section 11): fork `pack-loop` 05ef0fa3, `job-console`, the rest of `opencl-rdna4`, `opencl-rdna4-telemetry` | | unchanged |