1952 Commits

Include renames

Author	SHA1	Message	Date
jschoubben	ea53b2f920	mesh-build-go: each package counts its skipped tests (issue 890)	2026-10-26 12:29:19 +01:00
jschoubben	15a5fe09b6	mesh-build-go, mesh-build-node: run every package of a check's level after a failure (issue 1921) ... Pins mesh-sdk fix/1921-every-failure-r, whose check ask no longer stops a level at its first failing test step. A captured fixture with a failing test in each of two packages proves the build module's test step names both.	2026-10-26 10:58:07 +01:00
jschoubben	e7aaa8420e	Vault: name a hash record's secret in a field called of, so the runtime accepts it ... The runtime refuses a state field whose name ends in secret, so no bus-account hash could be written on novox. The test state now applies the same check. Issue: 1945	2026-10-26 09:25:35 +01:00
jschoubben	1c55e5d44d	Vault: take and hash a process's own bus account (broker-<process id>) as a covered account, so take-over is not held by every process user ... The controller hands each process account of a core or moved module with origin issued (ADR 0436 §2); the vault took only broker and provisioner-broker, refused these and never hashed them. Owed hashes are now said once per pass, and account_hash says who asked when it refuses too (review round 1). Issue: 1945	2026-10-26 02:35:57 +01:00
jschoubben	32b912cfe5	Vault: hash each bus account it is handed into bus-accounts.<user>, once per generation, so the grants part can compose the user list without the controller ... The controller still mints each covered account and hands it to the vault (ADR 0374); ADR 0547 clause 9 has the vault write the bcrypt hash of what it holds into its own state, readable by the grants part, and say account-hashed. It hashes on keep, at start every held account with no key of its generation, and again by hand (account_hash); a refused write is owed and written at the next pass. secret_own says per bus account whether this generation's hash is written. Issue: 1945	2026-10-26 02:28:28 +01:00
jschoubben	e572937d97	Vault: secret_own reads the module-own secrets it keeps, as fingerprints, so a generation can be told without the value ... Issue: 1945	2026-10-26 02:09:03 +01:00
jschoubben	e901769cb2	Read the newest steps first: sort the node's keys by id (start order) before the cut, and say older steps were not read ... The review found a cut read took an arbitrary 1000 keys, so the newest steps could be missing. A step id is base-36 milliseconds plus random digits, so its string order is its start order; the newest max(limit*4, 100) keys are read now, 1000 at most. The toolchains read logs when it drops keys past its cap.	2026-10-26 00:43:53 +01:00
jschoubben	8f9c78bcba	Read only the node's own steps, in parallel and bounded, so build_*_steps answers within the call's 30 s (issue 1827) ... On a steps state of 1164 keys / 20 MB (mesh-build-node: 10034 keys) the read tools listed every key and then read each value one round trip at a time, filtering by node only after the Get, so they timed out at 30 s. Now the keys are filtered by the node's prefix before any Get, the values are read 16 at a time, at most 1000 a call (the answer says it was cut), and one log line gives rows read and time. build_*_toolchains reads its values the same parallel way. The kind: a read tool that reads a whole state key by key; every read tool of the three modules is covered.	2026-10-26 00:39:33 +01:00
jschoubben	363a3d3f7e	Provisioner client: drop a module's held vault secrets when it leaves or is placed again, so it asks afresh (issue 1945) ... A re-assigned module was filled with the generation it held before the vault minted a new one; the bus refused the account.	2026-10-26 00:03:43 +01:00
jschoubben	32ad59fc64	mesh-delivery-gate and gitea: a setting check-gates-merge; false asks a reviewed pull request's Approve and merges it without waiting for mesh/check ... The operator, 2026-10-25 22:56: "run the check locally only, but not as a gate, it loses us too much time". Tonight each merge ask waited 3-7 minutes for mesh/check, and every merge into main refused the next approved merge for a check judged on the trunk before (mesh-delivery#52). Both settings default to true, so nothing changes until the operator sets them. False: the gate makes a reviewed pull request ready and asks its Approve with no verdict, a trunk's move puts nothing out of date, a verdict is kept and shown, and a group still merges in one pass; the merge tool merges a head the Approve names whatever its mesh/check says. delivery_status and show say the mode.	2026-10-25 23:07:50 +01:00
jschoubben	93f655dac9	mesh-delivery-gate: a merge learnt from the open list seconds after it landed moves its trunk and source.main; delivery_status shows source.main ... mesh-delivery#51 merged 21:43:11 and the gate read the open list at 21:43:14, before the forge's merged event: taken as learnt late, it moved no trunk and its group's finish left it out of source.main, so the checker kept judging every mesh-delivery head on 73be3df. A late merge within replayLate of its landing is now taken as heard as it happened, and one a gate before this left is written at the catch-up of the start (never backwards).	2026-10-25 23:01:37 +01:00
jschoubben	d0566f9ade	Fill a refused file again once the node-engine's declaration may let it through (ace, 2026-10-25 21:17) ... The node-engine refused mesh-build-go's files as not named as filled 32 s before its want arrived, and nothing filled them after the apply that named them: the runtime crash-looped on its missing broker credential. The client now fills every module again when the node-engine touches its write socket after such an apply (mesh-host fix/provisioner-retries-on-declaration), says each retry and each refusal it waits on, and asks the vault again at once for a refused secret when that wake or the module's own assignment may have ended the refusal.	2026-10-25 21:42:34 +01:00
jschoubben	9d9dfda19b	Merge pull request 'Gate: hand every member of an approved group in pin order, settle pull requests closed on the forge, and say a stopped pass as a condition (issue 1275)' (#480) from fix/1275-group-hands-every-member into main ... Reviewed-on: #480	2026-10-25 19:28:33 +00:00
jschoubben	4e82a824be	Announce the pauses state in the three build modules, and pin mesh-sdk 518dae5 (an unreadable pause pulls as not paused) ... Without it in Kept, stdio.KeptState refuses the state before the bus (mesh-sdk go/keeps.go refusedUnannounced), so the pause could be neither read nor written.	2026-10-25 18:57:08 +01:00
jschoubben	78d26867d2	Give the build modules' steps state an age of two days, and keep a node's pause in a pauses state of its own ... mesh-build-go's steps bucket filled to the bus's 64 MiB cap (27,370 keys, none ever removed) and refused the end of a passed check, so the check read failed; mesh-build-node's (17.8 MB) and mesh-build-image's grow the same way. A step's key is per step: its reader reads its end within the step's bound, and a check's verdict is kept by the checker, so two days is ample. The node's pause key lived in the same bucket and would have ended by itself; it moves to a pauses state without an age (mesh-sdk fix/steps-bucket-ages, Puller.PausePut). No node is paused today, so nothing is carried over.	2026-10-25 18:53:04 +01:00
jschoubben	0c216d99d0	Gate takes a merge learnt after the fact as past: no trunk move, the forge's time, never source.main (issue 1275, review round 2) ... A pull request found merged only by its absence from the open list went through PullMerged, which moved every other open delivery of its repository back to an old trunk (ready ones to proposed, composed checks out of date), dated the merge now, and so let a later catch-up write the old commit into source.main. pullMerged now takes such a merge as late: its own delivery is merged and its group closed, no other trunk moves, its time is the forge's or none, and LearntLate keeps it out of source.main in finish, settleForgeMerges and the catch-up. The gitea tool gitea_commit_statuses now answers the pull request's merged_at, and the gate reads a gone pull request by its number alone, so a deleted branch does not fail the read.	2026-10-25 18:37:59 +01:00
jschoubben	145c54d19c	Gate settles pull requests closed with no event, looks again at a line's bound, and keeps a waiting head's condition over a restart (issue 1275, review round 1) ... The conditions the gate now says would otherwise be wrong three ways: - A pull request closed or merged on the forge with no event heard stayed ready here for days (de-spiegel #4, closed 2026-10-17) and would raise a condition. After a complete read of the open list (shorter than the forge tool's 1000), each delivery not final whose pull request is missing is read from the forge and taken as its close or merge; a merge learnt so does not move source.main, as it is undated and could move it back. - A line that comes due on the clock (a state's bound, the author's bound, an answered refusal, an out-of-date refusal turning urgent after 10 minutes) waited for an unrelated event. The gate now asks a look at the earliest such time. - After a restart the heads waiting on the forge start their bound again, so their conditions were cleared and raised again later. A condition is cleared only once the open list was read since start, and not while the forge still shows its head open and waiting. Also: an empty path list ranks with a truncated one in the merge order, and a seat that keeps refusing conditions is retried on the clock 10 looks in a row, then at events only, said once in the log.	2026-10-25 18:32:19 +01:00
jschoubben	67092161bd	Gate hands every member of an approved group, in pin order, and says a stopped pass (issue 1275) ... Met again 2026-10-25: group fix/puller-wakes-on-capacity was merged half. Three causes in the gate: - The next member's turn compared the group's state with "delivering", a word GroupState no longer says since release 6 renamed it "merging" (ready.go groupTurnToMerge). Once the first member's merge was heard from the forge the group read merging, and the second member waited for ever. Group states are now named constants, the turn goes on in ready and merging (groupMayMerge), and a test holds every word to it. - Members merged in id order, so mesh-catalog went before the mesh-sdk commit it pins. membersOf now orders by mergeRank: mesh-sdk first, then members that change no go.mod or go.sum, then those that do. - No condition was raised: the controller's self-check that read `stalled` went with release 6 (mesh-controller dc456d56). The gate now says its stalled lines as conditions itself on the operator-channel seat (ADR 0437 §9) after every look, clears those listed no more, and keeps which it raised in its new bucket conditions. A held group's line says "half merged" when a member merged. A pass held for a required status no longer retries on a timer: the status turning success is the event that asks the merge again.	2026-10-25 18:20:58 +01:00
jschoubben	23c827e791	Take the mesh-sdk head that wakes a puller on its capacity, within the test budget, in the three build modules (issue 1887)	2026-10-25 17:44:50 +01:00
jschoubben	3138570c6b	Let the build modules follow their capacity on the bus, so a node raised from capacity 0 starts pulling (issue 1887) ... g14's Go builder sat idle at capacity 2 because nothing woke its parked puller. mesh-build-go, mesh-build-node and mesh-build-image now pin mesh-sdk b9911dc1, whose puller watches the config value capacity as mesh-config resolved it and wakes on each change, and hand it that watch.	2026-10-25 17:28:00 +01:00
jschoubben	1cc54959af	nextcloud: repair the account only on a 401, never reset the throttle (issue 1253, review 2) ... Resetting Nextcloud's brute-force record for the docker gateway cleared it for every client behind that address: someone guessing passwords could drive the throttle to 429 and have the module's poll wipe it. A throttle is no evidence the password is wrong, so a 429 is now backed off only, and the back-off outlasts the throttle. The account's existence is read from user:info's exit status, since user:list pages.	2026-10-25 16:57:46 +01:00
jschoubben	962bc359a7	nextcloud: repair only the module's own account, read its existence from the user list, no admin group (issue 1253, review 1) ... occ's user:info says 'user not found' on stdout with an empty stderr, so the error read as 'occ produced no output' and mesh-admin was never made on novox: occ now says stderr and stdout both, and existence is read from user:list. The repair could reset whatever account the user setting names, the operator's included, and the 'admin' fallback is a hand-made install's human administrator: only mesh-admin is repaired now, and it is the default. The shares poll needs no admin rights, so the account gets no group. After a repair the throttle of the caller's address is reset, and a failing repair reaches the back-off's one loud line with occ's message.	2026-10-25 16:54:51 +01:00
jschoubben	67b13c42d3	nextcloud: repair the account on a throttle too, since a throttle follows refusals (issue 1253) ... Right after the delivery the first answer may be Nextcloud's 429, not a 401; without a repair then the account would never be made and the back-off would wait out a throttle that cannot end.	2026-10-25 16:40:54 +01:00
jschoubben	4021724ca4	nextcloud: make the mesh's own account match its secret when Nextcloud refuses it, and back off a refusal that stays (issue 1253) ... The image creates NEXTCLOUD_ADMIN_USER only on a fresh install. novox's instance was installed before the mesh adopted it, so mesh-admin was never created there, and the shares poll asked Nextcloud every minute with a login that cannot pass: 401, then Nextcloud's brute-force throttle (429), which also held the backup before a delivery. On a refused login the client now adds the account through occ (admin group), or sets its password to the module's own secret, passing it by OC_PASS so it is on no argv, and asks once more. A refusal that stays is said once and the poll backs off from 5 minutes to an hour instead of nine lines of XML every minute. An empty setting no longer becomes an empty name.	2026-10-25 16:40:01 +01:00
jschoubben	155124b59a	mesh-build-go: a cause is the runtime's line at column 0, and only go's exact summary lines are dropped (issue 1901, review round 1) ... An indented log line of another running test that held 'panic:' won over the real timeout panic, since the lines were trimmed before matching. Go prints the runtime's lines at column 0; a test's log is indented.	2026-10-25 15:58:07 +01:00
jschoubben	6ec6fdaad2	mesh-build-go: a race the race detector reports is a failure's cause, in a test or after PASS (issue 1901) ... mesh-controller's try was worded 'PASS' again because the race detector reported after the tests passed and the program exited 66. The cause is now the race: its access, the frame that made it and where its goroutine was created, instead of PASS or testing's generic 'race detected during execution of test'. Tests from captured go test -race -json (1901/AC4).	2026-10-25 15:52:20 +01:00
jschoubben	663f3de975	mesh-build-go: a package that fails without an assertion carries its real cause, never "PASS" (issue 1901) ... A try of mesh-controller kept 'PASS' as its only failure: the package's test program stopped after printing PASS and the package-level failure took its first line. Now a failure that is not one test's assertion names the tests that were running and carries the panic line (a timeout's with its running tests), a signal or fatal error line, a build's first error, or words saying the program exited during which tests or after they passed; go test's summary lines are never kept. Tests from captured go test -json of eight real failures (1901/AC4).	2026-10-25 15:38:28 +01:00
jschoubben	53749833bf	Fill every module again when the node-engine's socket returns and when the node's own composition moves, so a file an apply left empty is filled (issue 1945)	2026-10-25 02:02:02 +02:00
jschoubben	ac6712cbb5	Pin mesh-sdk to feat/r6 748c564, which carries main's composer (issue 1943)	2026-10-25 00:02:45 +02:00
jschoubben	5b32342ac2	Merge main into feat/r6: keep R6's gate, which already carries #474's catch-up at start, and drop the seat and walk adoption R6 removed ... The reads-outside markers, mesh-build-go's compose job and gitea's forge-account fix come in from main.	2026-10-24 23:03:42 +02:00
jschoubben	6fb0ceeef8	Merge pull request 'R4g finish: mesh-build-go's compose job, the reads-outside marker in the digest, merge-check.sh gone (issues 1295, 1621-1623, 1935)' (#473) from feat/r4g-finish into main ... Reviewed-on: #473	2026-10-24 20:56:49 +00:00
mesh-tools	c651161787	Merge pull request 'gitea's forge-account reads the forge's own host key file before scanning its sshd, so a forge restart no longer refuses every forge account (issue 1939)' (#472) from fix/1939-forge-host-key into main	2026-10-24 20:05:39 +00:00
jschoubben	b67b5af5e6	Pin the gate to mesh-sdk feat/r6 at d7a9b968 (issue 1946)	2026-10-24 21:18:26 +02:00
jschoubben	d700ceca20	Pin the gate to mesh-sdk feat/r6 at 7f9cc0e5 (issue 1946)	2026-10-24 20:43:41 +02:00
jschoubben	3a1017f559	Pin the gate to mesh-sdk feat/r6 at e45bdc08 (issue 1946)	2026-10-24 20:22:07 +02:00
jschoubben	a17d0c8e93	The messenger holds and tells what the delivery view says is in motion, not the controller's plans (issue 1946)	2026-10-24 18:25:38 +02:00
jschoubben	9fa6710296	Pin the gate to mesh-sdk feat/r6 at a523a8cb (issue 1946)	2026-10-24 18:18:06 +02:00
jschoubben	05f849979c	Match a fix to a delivered revision by what the gate says of its merge; keep #474's source catch-up at start (issues 1946, 1726)	2026-10-24 18:13:40 +02:00
jschoubben	4dee77df23	Pin the gate to the head of mesh-sdk feat/r6 (issue 1946)	2026-10-24 17:54:30 +02:00
jschoubben	8a79877a15	A newer delivered revision covers the older merges of its repository, as the forge says ancestry (issue 1946)	2026-10-24 17:03:20 +02:00
jschoubben	2139e83a8a	Remove a binary built by mistake	2026-10-24 16:58:33 +02:00
jschoubben	e7ca8f7af1	The backup module says on the bus when a backup asked for now has ended, so no one polls backed-up (issue 1944, ADR 0539 §16)	2026-10-24 16:58:19 +02:00
jschoubben	eb79996336	The tracker derives a fix's delivered from the delivery view's event, not the gate's gone states (issue 1946)	2026-10-24 16:50:50 +02:00
jschoubben	1a271a1fb1	The group's integration test hears the forge's merges once both members are handed (issue 1946, R6-C3)	2026-10-24 16:43:48 +02:00
jschoubben	30d04f6bef	Pin the gate to mesh-sdk's release 6 verb tables and prove it holds the seat as they have it (issue 1946, R6-C3) ... The integration tests hear the forge's own word that a pull request merged, as the real flow does, and fail on any call of the controller.	2026-10-24 16:35:15 +02:00
jschoubben	5ff337e0dc	Gate keeps the merge's six states and asks the controller nothing (issue 1946, R6-C1, C2, C4) ... Release 6 deletes the plans: the walk half of the holder, the controller port, the carry-over of the former module and the verbs times, what-if, release, stop and retire-history go; the table holds proposed, ready, rejected, merged, held and closed; the delivery's status and view name the source revision the merge moved.	2026-10-24 16:26:36 +02:00
jschoubben	2fabbe556e	Pin to the group's new mesh-sdk head bfc3883a, whose composer reads a fallback to a predefined key as the indexer does (ADR 0348 §7)	2026-10-24 14:57:35 +02:00
jschoubben	89b9212c4d	Pin to the group's new mesh-sdk head 19c1d359, whose composer names the reader's pinned SDK instead of a bare (devel) (ADR 0482 clause 5)	2026-10-24 14:12:09 +02:00
jschoubben	40da368ee3	Date an adopted merge once, by the walk it was adopted from, so a later walk it follows never makes it look newer than a forge merge heard after it (issue 1726, review)	2026-10-24 13:55:10 +02:00
jschoubben	865f3b6455	Say what pacman-tools reads outside: the root predefined.json (issue 1935)	2026-10-24 13:48:18 +02:00