fleet/fleetd
3
0
Fork 0
Code Issues 73 Pull Requests Actions Packages Projects Releases 1 Wiki Activity

fleetd #422 follow-up: make the model gate's own state observable #434

Merged
ltms merged 1 commits from worker/422-gate-state-observability-9e79d6-11 into main 2026-09-10 08:32:55 +02:00
Conversation 0 Commits 1 Files Changed 7
+388 -10
agent commented 2026-09-10 08:19:16 +02:00
Member
Copy Link
Copy Source

Closes fleetd #422 follow-up.

Problem: PeerLauncher.disabledModels() returned an empty set both when a host has no models: block at all (nothing gated) and when a block exists with nothing currently off (gate armed, reporting zero). fleet_profiles/GET /profiles and the startup log could not tell the two apart.

Fix:

  • Added PeerLauncher.ModelGateState(configured, off) and a modelGateState() default method; disabledModels() now delegates to it, so no caller reads a different source.
  • CompositePeerLauncher.modelGateState() reads models0() once (the exact accessor the spawn gate itself reads) and distinguishes a null supplier (no models: block) from a real config-supplied block via identity against the private NO_MODELS_CONFIGURED sentinel.
  • New startup log line via Fleetd.modelGateCoverageLine, following the exhaustedPatternCoverageLine/errorPatternCoverageLine shape from #415.
  • New modelGateArmed field in FleetMcp.profilesView, reported unconditionally (never omitted) alongside the existing modelsOff set.

Deliberately did NOT reuse CompletionResolver.UnsetMeaning: that enum exists so two DIFFERENT keys (exhaustedPattern/errorPattern) can each state what their own empty-coverage means. Here there's only one gate and one accessor, so a small dedicated ModelGateState record (configured + off) was a better fit than threading an unrelated enum through it.

Tests:

  • FleetdModelGateCoverageLineTest (4 tests) — the three states plus pairwise-distinct wording, following FleetdPatternCoverageLineTest's shape.
  • FleetProfilesModelGateStateTest (3 tests) — asserts through FleetMcp.profilesView, the actual surface a lead reads.
  • CompositePeerLauncherTest: +2 tests — modelGateStateIsHotReloadedThroughARealConfigRef (real ConfigRef.reload() walking all three states live, no restart) and noModelsBlockConfigStillLoadsAndSpawnsNormally (proves the invariant: no models: block loads, validates, and still spawns).

Build: mvn clean install — Tests run: 1544, Failures: 0, Errors: 0, Skipped: 0. BUILD SUCCESS (main was at 1535; +9 new tests).

Mutation proof (see PR description / worker report for full table): 4 mutations run, each killing a distinct test or test set — including the two required ones (always-armed, ignoring no-block; always-not-configured, ignoring an off model), plus one each for the Fleetd and FleetMcp sites. All reverted; final build reconfirmed green after revert.

Closes fleetd #422 follow-up. **Problem:** PeerLauncher.disabledModels() returned an empty set both when a host has no models: block at all (nothing gated) and when a block exists with nothing currently off (gate armed, reporting zero). fleet_profiles/GET /profiles and the startup log could not tell the two apart. **Fix:** - Added PeerLauncher.ModelGateState(configured, off) and a modelGateState() default method; disabledModels() now delegates to it, so no caller reads a different source. - CompositePeerLauncher.modelGateState() reads models0() once (the exact accessor the spawn gate itself reads) and distinguishes a null supplier (no models: block) from a real config-supplied block via identity against the private NO_MODELS_CONFIGURED sentinel. - New startup log line via Fleetd.modelGateCoverageLine, following the exhaustedPatternCoverageLine/errorPatternCoverageLine shape from #415. - New modelGateArmed field in FleetMcp.profilesView, reported unconditionally (never omitted) alongside the existing modelsOff set. Deliberately did NOT reuse CompletionResolver.UnsetMeaning: that enum exists so two DIFFERENT keys (exhaustedPattern/errorPattern) can each state what their own empty-coverage means. Here there's only one gate and one accessor, so a small dedicated ModelGateState record (configured + off) was a better fit than threading an unrelated enum through it. **Tests:** - FleetdModelGateCoverageLineTest (4 tests) — the three states plus pairwise-distinct wording, following FleetdPatternCoverageLineTest's shape. - FleetProfilesModelGateStateTest (3 tests) — asserts through FleetMcp.profilesView, the actual surface a lead reads. - CompositePeerLauncherTest: +2 tests — modelGateStateIsHotReloadedThroughARealConfigRef (real ConfigRef.reload() walking all three states live, no restart) and noModelsBlockConfigStillLoadsAndSpawnsNormally (proves the invariant: no models: block loads, validates, and still spawns). **Build:** mvn clean install — Tests run: 1544, Failures: 0, Errors: 0, Skipped: 0. BUILD SUCCESS (main was at 1535; +9 new tests). **Mutation proof (see PR description / worker report for full table):** 4 mutations run, each killing a distinct test or test set — including the two required ones (always-armed, ignoring no-block; always-not-configured, ignoring an off model), plus one each for the Fleetd and FleetMcp sites. All reverted; final build reconfirmed green after revert.
agent added 1 commit 2026-09-10 08:19:17 +02:00
fleetd #422 follow-up: make the model gate's own state observable
CI / contract (pull_request) Successful in 45s
Details
CI / build (pull_request) Successful in 1m53s
Details
7fd914df1a 
PeerLauncher.disabledModels() reported an empty set both when no
models: block exists and when a block exists with nothing off, so
fleet_profiles/GET /profiles and the startup log could not tell an
inert gate from an armed one reporting zero. Add
PeerLauncher.ModelGateState (configured + off), a modelGateState()
default method disabledModels() now delegates to, and a
CompositePeerLauncher override that reads models0() once and
distinguishes the null-supplier case (no models: block) from a real,
config-supplied block via identity against the NO_MODELS_CONFIGURED
sentinel — reusing the exact accessor the spawn gate itself reads, per
the fleetd #404 lesson.

Wires the state into a new startup log line (Fleetd.modelGateCoverageLine)
and a new modelGateArmed field in FleetMcp.profilesView, reported
unconditionally alongside the existing modelsOff set.
ltms merged commit 6b7caba248 into main 2026-09-10 08:32:55 +02:00
Sign in to join this conversation.
Reviewers
No Reviewers
Labels
Clear labels
blocked
Cannot start until something else lands. The body says what.
 needs-live-proof
Merged and green, but never shown working on the running daemon. Not the same as done.
 ready-to-delegate
Scope, files and acceptance criteria are written. A worker can be briefed from the body alone.
 silent-default
A feature that compiles, passes tests, and ships turned off. Nine recurrences and counting.
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
agent kevin ltms
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: fleet/fleetd#434
Delete Branch "worker/422-gate-state-observability-9e79d6-11"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?