diff --git a/docs/plans/counter-asic-3-status.md b/docs/plans/counter-asic-3-status.md index 0ad3f9d8e..05668ff74 100644 --- a/docs/plans/counter-asic-3-status.md +++ b/docs/plans/counter-asic-3-status.md @@ -820,6 +820,28 @@ HIVE CANARY 4, RUN 1 PASS (the fleet lane, lp-4090-43): the re-cut package cf939 F0 SIGNED BY ALL THREE ROLES (read by the coordinator from the box mirror's master cd46db44b at 08:3x UK, all three rows present): the Protocol owner "the node lane (a283f5f0d364ceef0) | b792c8cc (F0's signing block as it stands; the manifest's head on master at signing b756ef26e, no identity pin corrected since) | 9 October 2026, 08:2x UK" (the merge of f0-sign-protocol2 d8392f5b, gate GREEN 92 checks; its note asserting the digest be5f4068 on devnet-4 (chain id 4465, no live move since 4cdcc488), the class v5 fingerprint cbc5bd0a with the hash lane's 08:27 line quoted and its recipe, the signing id 2a1d6caab4c24564 read equal by every reader, kit 2's 5d53a591 as the roll's binary (one source, three builds: fbf1d6e5, 28d0d4e5, 0d4680c5), c363ce90 as the 2.0.3 line; not convergence (kit 2 did not converge the fleet; lp-04 standing since 05:45:43; every read after that clock against a standing reference; 2.0.0's executor state not deterministic across nodes, 2.0.2's deterministic) and no PASS of a registry case; the block's pow-freeze.txt citation named as absent); the Product owner (the shipper, 81cfcdc7, 08:15); the Independent review owner (the adversary lane, f644a48a, 08:23). The hive class closed on the fleet lane's reads (the hive binary passing 3,686/3,687 through box-dn3.sh at 08:23:10 and under h-run.sh with the seven peers and the limit raised at 08:28:51); hive-4 run 2 names whether hive-1's wall was the peer set or the 1,024 limit. +HIVE-4 RUN 2 PASS AND HIVE-5 (the fleet lane, lp-4090-43): run 2, h-run.sh of the re-cut package, a fresh datadir, PEERS the three hubs only in the Flight Sheet config, the shell's limit raised (soft 1,024 before, 524,288 after; the node pid's /proc limits 524,288/524,288), started 08:30:19 UK: 3,686/3,687 VERIFIED first at 08:32:58 UK (168 s in, the same as run 1), not_held 0, 361 verifies by 08:33:34, blocks 8,037 of headers 10,796, the gate holding the miners; h-run and the node stopped by their recorded pids, left_node 0; so under the raised limit h-run's shape passes the records with hive-1's configured peers, where hive-1 walled at 04:03 UK under 1,024. Two honesty notes: (a) "the three hubs only" is the configured list; the node connected 10 peers by address gossip (h-run passes --nodnsseed but the address manager dials gossiped peers), as hive-1 did ("every other connected peer (7 in all)"), the same class, not strictly three; (b) hive-1 ran kit 2's hive package e475f3e0 without the sync gate, mining and voting from its first minute under 1,024 fds, while hive-3 and hive-4 did not mine during the sync: the gate a co-variable. Hive-5 from 08:35 UK on the same box: hive-1's exact package e475f3e0 by its own h-run.sh (no gate, mining from the start), PEERS the three hubs, the raised limit, 15 minutes (03-run-kit2pkg under hive-4/); a PASS names the limit, a FAIL the mining load under the old package; verdict by about 08:52. +THE h-run.sh FIX LANDED (the shipper, 08:38:35 UK): release-2.0.2 is c7f43510 (9998913f plus one commit to packaging/hive/h-run.sh only; the push gate GREEN, 60 checks in 76 s): at start ulimit -n 524288 best effort (else the hard limit), one line logging the soft limit before and after and the hard limit; the node's own /proc//limits "open files" line logged once it runs; the cleanup ending the node (its pid file), every miner (a per-GPU .gpuN.pid) and the loops (pids[]) by TERM then KILL on those pids, no pkill (kill-by-name-check clean; a stub test); the cutter's re-cut of the two hive packages from 5d53a591 and c7f43510 into /srv/artefacts/202-5d53a591-c7f43510/ by 09:30, read back on build-1; no entry today until the chain moves and a canary PASSes to a moving tip; the site lane's rebase target master 45c4c266, publish-202 rebasing on it before any publish. The landing hand: the adversary lane's morning delta (aaf3405b, section 14's routing-fault record) on master e3d0208d at 08:33 UK; thirty-five landings by the hand through it, each read back from the mirror. +MAIN'S WORD (08:4x UK): the founder is awake with the choice in front of him (go continue recommended, go rollback the other), so the 09:00 "go continue" default does not fire; the recipe waits on his exact word through main; founder-go.txt does not exist; nothing runs. +THE RECORD'S VERDICT AND THE FREEZE-FILE CITATION (the steward, master 1e8049af at 08:43 UK, ci-record-verdict 9c860585, gate green 92 checks, under the lock): kit 2's record tools/ci/canary/1176efb9.json reads verdict FAIL with its six blocks (mac PASS; fleet FAIL; hive FAIL; hive-2 FAIL; the kit-binary block FAIL against the standing reference; hive-3 FAIL with the sync gate; no windows block), the field set in the shipper's landing 21e327fb; the steward's landing adds the rule to tools/ci/canary-check.sh that a record whose own verdict reads PASS while any block reads FAIL is refused at the gate (self-tested); --artefact mac still reads PASS for the Mac entry's evidence; INT-07 FAIL unchanged. F0's mining-class row: packaging/pow-freeze.txt is the node fork's file (igneum-node, read by consensus/pow/build.rs under rule 19 at every kaspad build with the igneum-pow feature), never carried in the main tree; the row now cites igneum-node:packaging/pow-freeze.txt plus the rule-19 lines (the kaspad check with igneum-pow GREEN in each 2.0.2 pair batch's check:freeze cell, the freeze fingerprint cbc5bd0a read by build.rs's own method); a copy in this tree would be a mirror never the source, a ten-minute landing on the word if the protocol owner's signature needs it. GOV-01 and INT-07 moved with their pages; the steward's landings 20. + +THE FOUNDER'S WORD (08:44 UK, relayed by main, in founder-go.txt): "go continue". The operator reset by the founder's decision begins on the staged recipe morning-reset.sh sha fb746667c2d28ef6: the shipper stages the mini's datadir snapshot now (about 20 minutes: the mini's node stopped by its pid, the live devnet-4 datadir tarred with its sha, the app relaunched, the tarball pushed to build-1:/srv/artefacts/mini-dn4-snapshot/ and read back by sha); the fleet lane runs the recipe on its line (lp-05 and lp-06 first from the snapshot on the 5d53a591 kit, the wait for lp-04's tip 8c46c8a8 synced on both then lp-05's DAA past 46,800, lp-04 last from the same snapshot, then the fleet in groups with SEED the three hubs, lp-4090-43 last after hive-5); every line into reset-decision.txt; the island datadirs set aside, never deleted. Main's rule for the morning: the coordinator's 09:00 line opens on the reset's state (the snapshot staged or not, the hubs' stage started or not); after it, lines to main only at stage ends or faults. +THE SNAPSHOT STEP STARTED (the shipper): the founder's go read at 08:44:47 UK; the mini's go file written and ~/mini-snapshot.sh started at 08:44:52 under ~/mini-snapshot.pid (pid 3588): api/quit bounded, the host by its pid, the live devnet-4 datadir (7.5 GB) tarred with sha256 and size while the node is down (3 to 5 minutes), the app relaunched on the kept datadir at once (the three hubs env; it holds the 2.0.2 chain to 46,800 while the hubs come up), then the tarball and sidecar pushed to build-1:/srv/artefacts/mini-dn4-snapshot/ and sha256sum -c there (11 to 17 minutes at 7.7 MB/s); the one line with the path, sha and the restart time to the coordinator, the fleet lane and main the minute the box reads the sha back (about 09:05); the recipe's SEED lp-4090-43's node and build-1's hands (the mini not reachable). +THE RUNNER ARMED ON THE SNAPSHOT (the fleet lane, 08:4x UK): founder-go.txt holds exactly "go continue" (12 bytes, 08:44:30 UK); reset-on-snapshot.sh armed under pids/reset-on-snapshot.pid, "GO: ... waiting for the mini snapshot" in the decision file; it polls build-1 every 30 s for /srv/artefacts/mini-dn4-snapshot/.tgz with a non-empty sidecar and starts morning-reset.sh (fb746667c2d28ef6) under pids/morning-reset.pid the moment both exist (the recipe re-verifying the sha on the Mac, refusing on a mismatch); then the stages as staged (SNAPSHOT lp-4090-05, SNAPSHOT lp-4090-06, the 30-minute wait for both synced at or past 46,800, the 20-minute wait for lp-05's DAA past 46,800, SNAPSHOT lp-4090-04 last, then DIAL, HUBS2, MINERS with lp-4090-11, NODES, OUTAGE with every 2.0.1 miner and dn2-2, LAST lp-4090-43, RESET END, the within-ten tracker), the island count before and after each stage and every start line into reset-decision.txt, each stage's end relayed (the hubs past the cut with the first new block's height and time from lp-05's log, the fleet's one-tip read, the unreachable boxes by name), a wall its STOP line verbatim, never worked around. The clocks from the snapshot's landing: the hubs 15 to 20 minutes, one tip among the fleet about 55 minutes after that. Hive-5 on lp-4090-43 ends about 08:52 (that box last, about an hour later); the 2.0.3-k6 run on lp-4090-11 ends 08:56 (the MINERS stage after the hubs). Nothing touches a hub before the runner fires. +THE HIVE PACKAGES RE-CUT ON c7f43510 (the cutter, 08:45 UK, run pid 3609 08:38:57 to 08:45:25): /srv/artefacts/202-5d53a591-c7f43510/ holds the two hive packages from node 5d53a591 (the pair reused, igneumd 28d0d4e5) and app c7f43510: public igneum-hive-2.0.2-5d53a591.tar.gz sha256 ba08da7f658fcd969d2c7c2339eec81271dded3c069765523ca1c641d73c750c (52,852,859 B); lab igneum-hive-lab-2.0.2-5d53a591.tar.gz sha256 e79e51e56a85013b534cc716ab584c362a2223c22b49b2c2a91ed7e5765fe53b (52,852,875 B); both sidecars OK on the box; version.txt "igneumd 2.0.2-5d53a591" in both; h-run.sh inside both e640750aaa23c2e6ea50407331336b310ad5d9f7f2bfcd69038419631c0ba181, equal to the file at c7f43510 ("open files" 2, "sync gate" 7, pkill 0 in each); the narrow blake3 acceptance in run-hive-recut.log; the node tree untouched, no push, no manifest. +HIVE CANARY 6 (the fleet lane's default, the coordinator's silence as yes): the re-cut package c7f43510 (ba08da7f) pulled and verified on the Mac; from hive-5's end (about 08:52 UK) on lp-4090-43 by its own h-run.sh under a plain 1,024 shell (the package raising the limit itself, the thing to prove), PEERS the three hubs only, 15 minutes to the 3,686/3,687 verdict with its two limits and the quit by recorded pids read back, lines under build-1:/srv/canary/1176efb9/hive-6/; the box idle for the recipe's LAST stage after (about an hour later); the recipe's stage stopping it by pid files if it ever reached the box mid-run. +THE INTEL ARC GENERATOR 6 RE-RUN (the kit lane a6410f3b8abefb762, 08:5x UK): on PC 2 at 08:06 UK from the fresh dir and at 08:10 on zip 4 under its own id, both read back from the intake on build-1 (job-run-ca3-pc2-arc-g6-20261009 and job-run-ca3-pc2-arc-z4-20261009): the B580 reads the partner 6ce3dc344a613500 at 10.953 MH/s and the all pack e8f4f3289c6ee1fc at 5.465 MH/s, both PASS, identical across the two zips to three decimals, so the generator 6 all pack pairs on five of six platforms (CPU emulation, CUDA 4090, Metal, Apple OpenCL, Arc B580) and the partner on four; a tuning fact for the Intel lane: the all pack falls to the barrier exchange on the Arc (sub-group 16 for the 32-item group), the partner takes the shuffle path; the page row class-v5 9097d2c28. The 5090 and 7600 rows of zip 4 on PC 1 held by main's 08:48 order (the fleet install of 2.0.2 on PC 1 the moment its running job ends); the earliest clock the update-return lane's "PC 1 installed, mining on" line, then the fresh-id zip 4 fetch and the two kit reads through the hash lane's jobs with pc1-v6-bench.ps1 at 0c4f76ccf (a job id runs once per machine; a re-add under a used id ignored). +THE PC HOLD (the hash lane, 08:49 UK): the comparative job pc2-comparative-20261009 went live at 07:47Z before main's 08:48 hold reached the lane; it has not started, sits in PC 2's app queue (add-only, not withdrawable from the app); the relay lane may install first (the runner aborts and restores the card on the app's exit; the job republished under a new id after "PC 2 installed"); removed from the jobs file so nothing else fetches it; nothing new to PC 1 or PC 2 until the relay lane's installed lines; the kit lane and the comparative lane under the same hold. + +THE TWO 2.0.3 FLOWS ITEMS LANDED (the node lane, 08:49 UK, on main's 08:4x word that all three go onto the 2.0.3 line now, (1) and (2) with six suites green by 12:00): release-2.0.3-node-k6 = bf60948a on both node mirrors (the branch flows-2.0.3-ibd-and-close, the same two commits on c363ce90): 76cc5a88, the fd class: the router's outgoing sender sits in a slot that close() empties, so the transport's response stream ends on teardown whatever Arc lingers (enqueue reads ConnectionClosed), and the P2P server sets HTTP/2 keepalive (a ping every 10 s, closed after 20 s unanswered) and TCP keepalive, so a silent accepted connection is closed by the server; 13886a30's raised limit the delay it is; known-failed tests first (the lingering-sender stream in router.rs; the keepalive bound in connection_handler.rs). bf60948a, the standing-peer class: the IBD flow probes the peer's sink 5 s after the handshake and whenever this node's own sink has stood 300 s with no IBD running (under N4's 600 s), by the negotiation's own locator read; a sink held here as a header only or not at all, and not refused inside the N6 window, is handed to the peer's relay flow as the inv the peer never sent, the ordinary request/orphan/IBD path doing the rest; pure verdicts sink_probe_verdict and peer_sink_wants_ibd with known-failed tests. Six suites green on build-8 at 08:47 (p2p-lib 25, p2p-flows 44 (+2), miner 33, exec 102, consensus-core 186, consensus 146, pow 19). Next: the pair of bf60948a on build-2 and the one read showing item (1) without a moving chain (a fresh bf60948a node with --connect to lp-04 alone, which under c363ce90 sat at genesis for twelve minutes, must enter IBD within a minute), its line by 09:30. THE 09:15 WORD: (a) the steward's citation satisfies the protocol owner's signature (igneum-node:packaging/pow-freeze.txt read by consensus/pow/build.rs under rule 19, the check:freeze cell green in each 2.0.2 pair batch), no mirror, the note's correction reading "cited at igneum-node:packaging/pow-freeze.txt, read under rule 19"; (b) the order: (1) and (2) on the 2.0.3 line now; (3) the stale-sink relay read against the moving chain, then the HEAL lane's keyed scratch set, in that order on their own shas once the reset gives a chain; the hubs return on 5d53a591, bf60948a the upgrade after, canaried on the moving chain. + +HIVE-5: THE VARIABLE IS MINING DURING THE SYNC, NOT THE LIMIT (the fleet lane, lp-4090-43): hive-1's exact package e475f3e0 (kit 2's hive, no sync gate, the miners and voting from the first minute) by its own h-run.sh, PEERS the three hubs, the shell's limit raised (1,024 to 524,288; the node pid 277926 read back 524,288/524,288), started 08:35:08 UK: in 900 s no VERIFIED of 3,686/3,687 and no "not held yet"; the node reported synced=true at blocks 3,916 of headers 7,197 (its own island from the first cut at 3,600, growing at its own mining rate while the headers stood), never reaching the 3,687 band of lp-04's chain, mining on its own fork below it; everything stopped by recorded pids. Read against the night: hive-1 (the same package, 1,024) reached the band and walled on the lost proofs; hive-5 (the same package, 524,288) did not reach it; hive-3 (gated, 1,024) and hive-4 twice (gated, 524,288) passed the band in 168 s. So the open-files limit is not the variable; mining during the sync is: the ungated package starts its eight identities voting and mining on the first template the unsynced node hands out, the node islands itself, and the executor's state wait walls it at the first cut; the sync gate (9998913f and c7f43510) removes that; the raised limit a separate good for the fd class. The lines under build-1:/srv/canary/1176efb9/hive-4/03-run-kit2pkg. Hive-6 (the c7f43510 package: the gate plus h-run raising the limit itself, under a plain shell, the three hubs) from 08:50:49 UK, verdict about 09:06, the last run on that box before the recipe's LAST stage. + +TWO FOUNDER WORDS (09:0x UK, relayed by main): (1) hub-1's devnet-3 aside is deleted after its compressed archive is on build-1 with a sha list read back equal (the fleet lane's order, by 11:00, beside the reset). (2) The prover's blake3-dispatched AVX-512 lines: the founder's word is "do it if it is in the 2.0 plan"; the master pins the x86-64-v3 fleet baseline and the rule-33 canary proving a shard on a non-AVX-512 box, not a blake3-pure build, so rule 37 stands (the pinned served prover 71bc2438's 2,542 lines inside blake3_hash_many_avx512 and blake3_xof_many_avx512 accepted by named symbol) and no prover rebuild runs. +THE bf60948a EVIDENCE KIT (the cutter, 08:52 UK, run pid 10299): /srv/workers/fleet/bf60948a-node-lane.tgz sha256 be4eeb31f536554dcad1aeb92d19dc11d362ef5627587721e7f916b6025eeb7f, igneumd d88ba694799d9f4c "igneumd 2.0.3" with the commit string, miner 727606c025ffb0d3, fingerprint cbc5bd0aa10585c8 in both, kit-isa 0, no lab string, record 202-evidence-bf60948a; supersedes c363ce90 as the 2.0.3 canary candidate; nothing packaged or pushed. +THE hub-1 ARCHIVE ORDER (the fleet lane, 08:54 UK): the in-place compression pass stopped by its pid file (hub-1 free 5.2 GB; the aside 13,039 files, about 19 GB, partly compressed); the sha256 list of every file made on hub-1 (/root/fleet/dn3-aside.sha256, 13,039 lines) and copied to the Mac; the stream path hub-1 to the Mac to build-1 over two ssh pipes (hub-1 holds no key, by the rule). The block: build-1 has no /srv/archive and the build user cannot create it (root-owned /srv, no passwordless sudo), so the first attempt's tar had nowhere to land, its verify failed by construction and the guard held: nothing removed on hub-1 (the "NOT REMOVED" line in the decision file is that attempt); the build-server lane asked at 08:54 to create /srv/archive/hub-1-devnet-3 owned by build; the default if not there by 09:20 UK (the coordinator's silence as yes): the archive under build-1:/srv/artefacts/archive/hub-1-devnet-3/ (build-writable, 2.1 TB free) with the sha list beside it, the path in the done line for root to move later; the deletion on hub-1 only after every sha verifies equal, under the pid file, the free GB before and after in the decision file and one line for main; the clock: about 30 to 45 minutes at a 50 Mbit/s home link for about 10 GB compressed, done about 10:00 to 10:30 UK inside the 11:00 default; nothing else on hub-1 touched. +THE RESET'S STATE AT 08:55 UK (read by the coordinator): founder-go.txt "go continue"; the mini's snapshot streaming to build-1:/srv/artefacts/mini-dn4-snapshot/mini-dn4-snapshot-20261009-074452Z.tgz (4,567,511,040 B at 08:55, the sidecar not yet written), the runner polling for the sidecar; the hubs' stage not started; the feed standing at 24,852 / 8c46c8a8 / DAA 46,800; master 7e019fde0; kit 2's record's top-level verdict FAIL; the hub-1 archive's second attempt started 07:54:39Z (the aside 13,039 files, 12,744 MB; hub-1 free 6,596 MB; the sha list on build-1; the stream hub-1 to the Mac to build-1). + A SHARED-DEVNET FACT FROM THE FLEET (not this lane's, with the shipper and the infra lane): the Hetzner live seed 188.245.5.161:26611 is still on the old override object (digest eada4bda) 1 h 40 min after the 0.3.20 sweep (the fleet never touches Hetzner nodes, so it was outside the sweep); the 0.3.21 wipe canary c22-1 took five digest-mismatch rejects from it; an app with the packaged peers is refused at the seed and syncs through node1 and the hub only, a fresh joiner with only the seed cannot join, the 14 voters and the hub are unaffected; the owner puts the floor file ov16-floor-900000.json (sha 294f1f80) and the c4459193 pin on it. 0.3.21's STAGING (the node lane): the order dry-merges onto 55768f88 with nothing moving to 0.3.22; the late-join fix is 52e96c94 (70e4601e rebased onto 55768f88, exec suite 33 green with both new tests); f067f7c1, b0444f51 and 437f0438 merge clean in order; 2e32d5f6's one conflict (DST_ADDRESS beside pool-finish's DST_BINDING in consensus/core/src/finality.rs) kept both; the live-file digest eada4bda after each (every switch at never); the staging waits on the shipper's sweep-end word; the re-pin held. PC 2 DOWN AGAIN (main, 16:5x UK): the founder takes PC 2 down for cable work (PC 1 back but his desk); both PCs out of the sweep's waves, each updates on its poller on return; no PC job to PC 1; the Windows G1 completed before the outage, nothing reruns. 0.3.21's SECOND GATE LINE on 55768f88 (sha256 279b1b690e854fc9): the ten-minute mixed-version gate beside the 5899f603 pair, 13:37:40Z to 13:47:52Z, SUMMARY PASS (one digest b0afb2ee on five nodes; 223 new and 381 old blocks accepted by the old hub, 0 rejected; counts equal at 319, 486 and 604 through both clean joins and the restart step at 13:45:22Z; no panic); the node lane's two lines on 0.3.21's first candidate complete, in plan 6.9 on ca3-v4-node; the fleet's set on it (the bare-child 12 GB line, the wipe, the kept read, the cases) is the fleet's. 0.3.21's FIRST GATE LINE on 55768f88 (sha256 279b1b690e854fc9, the string read back; pairing igneum-pow 8c728ca3 at byte 5): the digest gate 13:35:41Z to 13:37:19Z SUMMARY PASS (a89be8a7 on both binaries with the peers; db9a85f9 refused, no peer; the live file's eada4bda unmoved); the ten-minute mixed-version gate from 13:37:40Z, line about 13:50Z. The 0.3.21 order as the shipper sent it: 55768f88; f067f7c1 and 70e4601e; b0444f51; 6eb21fc9; db28d331; then the re-pin from 8bdcbdd8 on the coordinator's word; suites between, the digest read after every one; the mirror's release-0.3.20-node back at the pin c4459193, release-0.3.21-node open at 55768f88. THE LATE-JOIN COMMIT (N9's second half, the node lane): 70e4601e on the box mirror as branch proof-hold-fix, from c4459193, two files (igneum/exec/src/proving.rs, protocol/flows/src/v10/proving.rs); the gap was the fetch side on the joiner (the served record ran the native check against the joiner's trailing exec state before anything was stored, the check refused it, the proof was never held, the body rule read "not held" for 20 s and failed the IBD); the fix holds the proof by hash before the checks (the pool entry still needs them) and the serve side says when it holds fewer than asked; the exec suite 32 passed at 13:26Z with the known-failed shape first, the flows check green 13:28Z, igneumd on build-1 at the 0321 worktree path built 13:32Z, sha256 17649eeb2f7d1290, string read back; with the testnet lane (the resume form, B alone); it joins the 0.3.21 staging as its own commit. THE WIPE CANARY ON c19-1, c4459193 (sha 45be9b02d1b002f5, string read back): FORM END rc 0 at 13:50:53Z. Wipe synced 13:35:50Z (57 minutes, inside the 98-minute class); mining 13:36:00Z to 13:47:07Z, 66 mined, 66 accepted, 0 rejected, isSynced true at the tip throughout; the hub holds 41 of its blocks in its last 700 with 0 rejects (13:47:09Z); the restart on its kept datadir at 13:47:15Z: the old process stopped at once (the new process's first lock line seven seconds after the marker; the watchdog held nothing, the b7cc37e7 fault closed), synced again at 13:48:39Z after 84 s, 109 templates read with max 3,432 ms and 0 timeouts; the kept read on pool-1's 0.3.17 copy on the same pod passed at 13:38Z (the rewrite line once, a clean second start). The pin's set on c4459193: the digest gate PASS, the mixed-version gate PASS, the wipe canary PASS, the kept read PASS, the restart PASS, the 12 GB line proves and verifies (paid is a race, not a gate); CASES END from c20-1 (about 14:50Z) is the last pin line. THE INTEROP FACT stands from the void run: the 5899f603 hub accepted 235 object-byte-5 blocks from the 8097d600 node with 0 rejected, one digest on all five nodes on the live sixteen-field file. The gates: the digest test and the kaspa-pow vector test (the amended devnet epoch-0 id 1a4230699a6b9c60 must equal, c120d7963abdcd96 must differ, the v3 control unchanged) on the box; the mixed-version Devnet 2 gate (the amended 0.3.20 node beside a 5899f603 node for ten minutes on the live file without the v4 fields) after the Mac build; the fresh-join canary the 0.3.20 cut's | | Main's rulings (7 October, morning) | no generator change to v4 on the live devnet; the record's null is the window model with numbers, sent by the hash lane to the attack-pass lane so AP-F8-1 re-gates against it; a fault beyond the model (a low-entropy source at site 15) stops at the coordinator with the two options priced (a 0.3.19 class amendment before the flip, or the flip held at the floor), nothing shipping without the founder's word; the tighter tail, an acceptance bound on the hot-set share, is a CLASS V5 item (sent to the v5 lane a6410f3b8abefb762 with the 64-seed census as its gate; the bound's number follows from the model) |