The check of 2026-11-02 22:35 on a loaded shanks failed two buildrun tests on wall-clock bounds of 60 ms, the
second failure of this kind tonight (the first: mesh-sdk's busretry tests, f02f142).
Quick fix: give the two buildrun asker tests bounds of seconds.
Thorough fix (chosen, because the kind recurs across every module of the repository): every test that raced a
wall clock now waits on an event, steps a clock it is given, or has a bound only a failing test reaches.
- buildrun asker tests: the steps' own times stand for the clock (bound an hour, starts and phases written in the
past), no goroutine paces phases with sleeps.
- builder: the decline bound reads Builder.Now; the decline test steps it; the stream-wait test waits for five
check reads instead of 100 ms.
- checker: the lost-ask try bounds an untaken ask at 200 ms only where that is what ends it; the program says when a
merge turn changes, and the integration test waits on that line instead of 300 ms.
- teller: the bed makes the teller's own ends bucket (a 10 s stall at start the tests raced by 0.5 s), waits for its
warrant subscription instead of 500 ms pauses, and the package-level now is restored with t.Cleanup.
- watchman: the recount, run in its own goroutine, is waited for instead of 10 ms.
- polls of 2-5 s become 10 s (unit) and 20 s (integration).
Under a throttled cgroup (half a core shared with two busy loops) beside the node-engine's suite, the buildrun
tests failed 5 of 20 runs before and passed 30 of 30 after; no test added.
Claude-Session: https://claude.ai/code/session_01NcbgfJ9Gum47zkFKWRaovL
The teller read the router's record from KV_asks, a bucket no composed grant
names, so every read was denied and its operator answers could not be read
back. It now reads messenger_asks, as mesh-grants does. The integration bed
grants the real subject and reads a record on the teller's credential, so it
no longer agrees with the mistake. (issue 774)
Claude-Session: https://claude.ai/code/session_01NcbgfJ9Gum47zkFKWRaovL
mesh/check checking 17abe6af32ce on af0fd83aec09 | novox/mesh-delivery#124 in group fix/merge-ask-ends-on-merge
On 2026-10-31 two merge asks of pull requests merged by another path held two of
the router's three places for the teller and a placement ask was refused for want
of room. The gate's fix ends such asks by itself; this tool is the recovery for any
ask that still outlives its question: it cancels the open ask at the router, or
drops one waiting for room, by its router id or its key, and logs who and why. The
tools bundle asks through the teller's new state ends, as indexer_take_in does.
mesh/check pass 4628ba26b41a on 73be3df28cd4 | unit 167 run, 443 kept | integration 177 run, 509 kept | composition passed on 3 machines
The open helpers were copied into seven parts; code shared between
modules lives in mesh-sdk, so they are mesh-sdk process's now and the
parts are pinned at fix/1945-optional-buckets (e0d1de04), moved to its
main once it merges, one delivery group with mesh-sdk.
Met twice on 2026-10-25: the builder of #48 exited on a refused read of
the gate's source and crash-looped, and its new reasks bucket would have
done the same. A grant reaches the bus with a delivery, so a refusal or a
timeout at start is often only a grant not there yet; exiting for it
stopped the whole part until a hand step on the bus.
Every startup open in the parts is now essential or optional. Essential
ones (a part's own state and the input it exists for) still stop the
start, with an error naming the bucket and its grant. Optional ones are
said and left: watches the bus refuses too. The builder builds trunk keys
without the gate's source, which it opens again at the next candidate
that asks a trunk head; toolchains and previews open again at their next
event, as a missing bucket did. builder_ask_again says that asking a
running step again builds it twice.
A hold answered keep on the phone could not be changed: its ask ended with the
answer, nothing listed it, and a kept hold is never lifted by itself. The view
lists every hold with its node, cause and answer (mesh-delivery-cli holds) and
answers the standing hold ask with hold-release or hold-keep through its one
write; the teller takes a later view answer that changes a hold's option into
the answer of that asking, never an older one.
A hold written again after a lift and a restart reused its first ask id, which
the router drops; the hold's time is now its asking. Seven holds could take the
router's three places for the teller and keep a merge Approve from being asked;
the teller keeps at most two assigner asks open. A release or keep answered for
one hold carried over to a later hold of the same fingerprint; an answer now
counts only for the asking its warrant names.
Seven modules stayed held on every machine and the operator was never asked:
the teller only filled assigner asks from the view, a single failed check held
a fingerprint for good, and the held record kept no cause. The teller now asks
every assigner ask as it asks merge asks and writes the phone's answer under
the ask's key; the assigner lifts a hold whose fingerprint runs healthy on
every node that runs the module, says so once and takes its ask back; the held
record names health, phase, when and the node's words.
mesh/merge-gate head 037bf1ef361b merged into trunk b04ee764739a — pass: builds mesh-delivery-teller → novox; no bus step; every machine composes with the …
The router holds at most three open asks of one asker and refuses a fourth without keeping it; the teller only logged the refusal, so a merge ask in a burst never reached the operator while the gate read it as asked (issues 1610, 1620). Bounded at six refusals, then said as a condition.
mesh/repo-check head 8e6a31bc6ab4 merged into trunk 7f618177d677 beside mesh-catalog#374@75957fc3 — pass: THE CHANGE ALTERS ITS OWN CHECK (merge-check.sh):…
mesh/merge-gate head 8e6a31bc6ab4 merged into trunk 7f618177d677 — pass: group feat/delivery-parts composed: every machine composes with the change as it d…
jschoubben
mesh-tools