27 Commits
Author SHA1 Message Date
kyle 7a12ddcf5a Context refusal in llama-server's exceed_context_size_error shape; README for v2.3
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 20:15:49 -07:00
kyle fa1c398fc4 A route may have its own listener: every request there is that route, paths unprefixed
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 19:59:21 -07:00
kyleandClaude Opus 5.5 aea2eeae2c Routes may share one lease (affinity = "route") and skip crossbar's queue (queue = false)
route.go gains Route.Affinity/Queue with PerRoute(), Queues() and affinity
validation (checkRoutes moved here; config.go calls it once). limiter.Track
counts a request without holding or refusing it; a release hands the slot to a
waiter only while in flight <= parallel. The proxy leases a PerRoute() route
under an empty fingerprint (the row keeps the real one) and uses Track when
Queues() is false.

Implemented by Ornith (OpenCode); owner review removed a release-on-first-flush
workaround for a race in the owner's given test (see implementer log).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-09-25 19:33:46 -07:00
kyle 33fa61bedb Control-plane requests follow the lease but take no slot and write no row
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 17:44:00 -07:00
kyle 8018f67031 Wake: a target may list several broadcast addresses
Config keeps broadcast (one) and adds broadcasts (a list); exactly one must
be present. Wake.Addresses() returns Broadcast then Broadcasts; checkWake
errors on both-set, neither-or-empty, and non-host:port entries. Waker sends
to every address in order, logging/past a failure so one dead address does
not stop the others, and returns false only when none could be sent.

Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 13:45:39 -07:00
kyle 4e1dd03d07 Route templates: a route named x-* serves any request route x-<something>
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 13:38:24 -07:00
kyle 055ab079d4 Learn per-model context from /props?model=; only status "loaded" is loaded
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 11:47:35 -07:00
kyle f3dfdbfa50 Record cancellation from what the reverse proxy observed, not the request context
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 11:32:18 -07:00
kyleandClaude Fable 5.1 98faa3a57d Given tests: order arrivals by limiter state instead of sleeps (limiter, spread, cancel-while-queued)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-09-25 11:20:26 -07:00
kyle d0d3203f73 Wire wake and identity into crossbar; v2 smoke and README
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 11:12:24 -07:00
kyle a942e336d8 Add identity: whois resolver, checker, header mode, middleware; config for wake, peers, identity
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 10:16:51 -07:00
kyle bddf6c93f2 Add the wake package: magic packets and a waiter
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 10:03:07 -07:00
kyle 2fe3b9c865 Proxy: move or refuse prompts that do not fit the leased host's context
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 09:56:46 -07:00
kyle 056a527fd7 Health: learn n_ctx and total_slots from /props
Implemented /props learning in internal/health/health.go (added Status.NCtx/Status.Slots, PerSlotCtx, and a best-effort GET <base>/props appended to the poll after /v1/models; 0/unknown on any failure without failing the poll) and exposed them in internal/admin/admin.go HostView. Copied internal/health/props_test.go and the replacement internal/proxy/helpers_test.go byte-identical to docs/plans/v2/_files/.

Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 09:14:08 -07:00
kyle 9f5b50acf6 Review fixes: record cancelled requests as 499; empty usage is an array
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 09:00:31 -07:00
kyle cf2aa24393 Smoke run for v1; README for leases, admin and accounting
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 06:56:54 -07:00
kyle 32ac7f549a Admin: leases, pin, release, drain, usage, metrics
cmd/crossbar/main.go calls admin.Handler with the new 6-arg signature, passing nil for the not-yet-wired leases/limiter/store/drainer (task 07 wires them) so go vet and go test ./... pass on cmd/crossbar. This is a compile fix, not the task-07 wiring; noted in the implementer-log.

Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 06:33:29 -07:00
kyle 97f7cdffd9 Route by lease, queue per host and model, record every request
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 05:15:21 -07:00
kyle 9133240cfb Add the sticky lease table
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 04:26:47 -07:00
kyle 7e0dbb4a6f Add the per-host-model limiter and the host chooser
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 03:53:38 -07:00
kyle 463cea18de Add the conversation fingerprint and the v1 config keys
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 03:46:05 -07:00
kyle 816614d6dd Add the SQLite store for leases and accounting
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 03:34:24 -07:00
kyle 9e0f906c8b Review fixes: recorder Flush without panic; unreadable config file is an error
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 02:58:17 -07:00
kyle 4bc5115f24 Add the admin endpoints, the crossbar binary and the fake upstream
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 02:32:50 -07:00
kyle 5b347d9dd9 Add the routing reverse proxy
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 02:26:19 -07:00
kyle 68ad94d693 Add the health poller
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 02:16:38 -07:00
kyle 73b24357dd Add the Go module, the gate and the config package
Implemented-By: OpenCode session (model recorded in docs/implementer-log.md)
2026-09-25 02:09:42 -07:00