diff --git a/.wolf/buglog.json b/.wolf/buglog.json index 954369c5..cadb27ea 100644 --- a/.wolf/buglog.json +++ b/.wolf/buglog.json @@ -18425,9 +18425,9 @@ }, { "id": "bug-1958", - "status": "open", + "status": "fixed", "status_updated": "2026-10-10", - "status_note": "newly logged", + "status_note": "", "component": "apiserver", "severity": "security", "timestamp": "2026-10-10T10:15:56Z", @@ -18435,8 +18435,8 @@ "occurrences": 1, "error_message": "apiserver lets any rbac:control grant create roles and extend its own role (privilege escalation)\n\n**Observed:** (source reading; follow-up of the 2026-10-10 P0 fixes (bug-1936..1957))\nReported by the bug-1952 fix: role create/update in `controllers/roles.go` accepts any `rbac:control` scope, so a user holding `rbac:control` with scope `self` can add `cluster:control` (or any permission) to a role it already holds: the same escalation bug-1952 closed for user-role assignment, by another route.\n\n**Expected:** Creating roles or changing a role's permissions requires `rbac:control` with scope `all` (as user-role changes do since bug-1952).\n**Impact:** Security: privilege escalation through role editing.\n**Found by:** follow-up of the 2026-10-10 P0 fixes (bug-1936..1957)", "file": "src/frontend/apiserver/controllers/roles.go", - "root_cause": "TODO: required for fixed/partial", - "fix": "TODO: required for fixed/partial\n\nVerification:\n- Test: TODO: name + file; RED before the fix, GREEN after\n- Suites: TODO: suite results and build profile\n- Live: TODO: soak/deploy run, or n/a and why", + "root_cause": "`CreateRole`, `AddPermissionsForRole` and `DeleteRoleAndPermissions` (`controllers/roles.go`) were guarded only by the route's `RequirePermission(\"rbac:control\")`, which accepts any scope, so a `self`-scoped grant could add `cluster:control` to a role its holder has. Pre-existing.", + "fix": "New `checkRoleChangePermission()` in `roles.go`, called first in the three handlers, reuses `canChangeUserRoles` (bug-1952; all grants before deduplication) and answers 403 unless the caller holds `rbac:control` with scope `all`.\n\nVerification:\n- Test: `TestAddPermissionsForRole_SelfScopeCannotExtendOwnRole` (`controllers/roles_rbac_test.go`): RED `roles_rbac_test.go:65: privilege escalation: mallory (rbac:control scope self) added cluster:control to its own role (status 200)`; `TestCreateRole_SelfScopeRejected`, `TestDeleteRole_SelfScopeRejected` RED (status 200, want 403); GREEN after, plus `TestRoleChanges_AllScopeAllowed`.\n- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router)\n- Live: n/a: unit/fixture-level verification; no live deployment was run", "tags": [ "apiserver", "security", @@ -18445,16 +18445,18 @@ "related_bugs": [ "bug-1952" ], - "fix_commits": [], + "fix_commits": [ + "736ed81d" + ], "record": ".wolf/bugs/bug-1958.md", "gitea_issue": 389, "gitea_synced_status": "open" }, { "id": "bug-1959", - "status": "open", + "status": "fixed", "status_updated": "2026-10-10", - "status_note": "newly logged", + "status_note": "", "component": "xcecli", "severity": "security", "timestamp": "2026-10-10T10:15:59Z", @@ -18462,8 +18464,8 @@ "occurrences": 1, "error_message": "xcecli logout only clears local config and never calls the logout API, so the token is not revoked\n\n**Observed:** (source reading; follow-up of the 2026-10-10 P0 fixes (bug-1936..1957))\nReported by the bug-1946 fix: server-side token revocation exists (`POST /xce/v1/auth/logout`), but `xcecli logout` deletes the local token without calling it, so the token stays valid until expiry.\n\n**Expected:** `xcecli logout` calls the logout API (best effort) before clearing local state.\n**Impact:** Security: a CLI token survives logout.\n**Found by:** follow-up of the 2026-10-10 P0 fixes (bug-1936..1957)", "file": "src/frontend/xcecli/cmd/auth", - "root_cause": "TODO: required for fixed/partial", - "fix": "TODO: required for fixed/partial\n\nVerification:\n- Test: TODO: name + file; RED before the fix, GREEN after\n- Suites: TODO: suite results and build profile\n- Live: TODO: soak/deploy run, or n/a and why", + "root_cause": "`runLogout` (`src/frontend/xcecli/cmd/auth/logout.go`) only removed the server entry from the local config; the CLI never called the logout API, so the token stayed valid server side until expiry. Pre-existing.", + "fix": "New `APIClient.Logout(token)` (`common/request/request.go`) posts `/xce/v1/auth/logout` with the bearer token; `runLogout` calls it first and, if the server is unreachable or rejects the token, warns `server-side token revocation could not be confirmed` and still clears local state; help text updated.\n\nVerification:\n- Test: `TestRunLogout_RevokesTokenOnServer` (`cmd/auth/logout_test.go`, httptest): RED `logout_test.go:124: requests = [], want [\"POST /xce/v1/auth/logout Bearer tok\"]`; `TestRunLogout_ServerFailureStillClearsLocalState` (rejected and unreachable server) RED `logout_test.go:157`; GREEN after (5 auth tests).\n- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router)\n- Live: n/a: unit/fixture-level verification; no live deployment was run", "tags": [ "xcecli", "security", @@ -18472,16 +18474,18 @@ "related_bugs": [ "bug-1946" ], - "fix_commits": [], + "fix_commits": [ + "03798230" + ], "record": ".wolf/bugs/bug-1959.md", "gitea_issue": 390, "gitea_synced_status": "open" }, { "id": "bug-1960", - "status": "open", + "status": "fixed", "status_updated": "2026-10-10", - "status_note": "newly logged", + "status_note": "", "component": "controller", "severity": "wrong-result", "timestamp": "2026-10-10T10:16:02Z", @@ -18489,8 +18493,8 @@ "occurrences": 1, "error_message": "closed queues accept submissions: controller never checks, apiserver skips the default queue\n\n**Observed:** (source reading; follow-up of the 2026-10-10 P0 fixes (bug-1936..1957))\nReported by the bug-1938 fix: `handle_jobsubmit()` (`src/controller/job.c`) never checks `QUEUE_STAT_CLOSED`; the apiserver's check in `preCheckJobRequest()` runs only when a queue name is given, so a submission without one goes to `normal` unchecked, and non-REST submit paths (libxce, harness) are not checked at all.\n\n**Expected:** A submission to a closed queue (explicit or default) is rejected with a clear error on every submit path.\n**Impact:** Admin queue close silently ignored for default-queue and non-REST submissions.\n**Found by:** follow-up of the 2026-10-10 P0 fixes (bug-1936..1957)", "file": "src/controller/job.c, src/frontend/apiserver/controllers/jobs.go", - "root_cause": "TODO: required for fixed/partial", - "fix": "TODO: required for fixed/partial\n\nVerification:\n- Test: TODO: name + file; RED before the fix, GREEN after\n- Suites: TODO: suite results and build profile\n- Live: TODO: soak/deploy run, or n/a and why", + "root_cause": "`handle_jobsubmit()` (`src/controller/job.c`) only checked that the queue field was non-empty, never the queue's existence or `QUEUE_STAT_CLOSED`, so single and array submissions from libxce/harness (and REST without a queue name) reached closed or unknown queues; `preCheckJobRequest()` (`controllers/jobs.go`) checked only a named queue. Pre-existing.", + "fix": "Controller (`b239db5f`): new `jobsubmit_check_queue()` (in-memory read under the qmgr/queue read locks) rejects an unknown queue or `lost_and_found` with `-E_XCE_BAD_QUEUE` and a closed queue with new `-E_XCE_QUEUE_CLOSED` (10016, `include/xce/errnox.h`, string in `lib/utils.c`) before the hook, the in-flight registry and any WAL append; replay/apply paths unchanged, so committed submits still apply. Apiserver (`1ff068f6`): the default queue gets the same existence/closed checks. Apiserver (`4f579041`): `SubmitJob` maps the two codes to `backend.ErrQueueClosed`/`ErrQueueInvalid` (checked against errnox.h at compile time) and the submit/restart handlers answer 400 naming the queue (covers a close racing the apiserver check). Behaviour change: a submission without a queue is rejected when the cluster has no `normal` queue.\n\nVerification:\n- Test: Controller: `test_jobsubmit_closed_queue_rejected`, `test_jobsubmit_unknown_queue_rejected`, `test_jobsubmit_lost_and_found_rejected` (`src/controller/testsuite/test_jobsubmit_validate.c`): RED `test_jobsubmit_validate.c:370: FAIL: Expected -10016 Was -5` (and -10008), GREEN after; `test_ace_job_submit_closed_queue_still_applies` guards replay. Apiserver: `jobs_test.go:137: submission to the closed default queue: err = ` RED, GREEN after; `TestSubmitRCError`, `TestJobHandler_SubmitQueueRejectedByController` failed to build before (no typed errors), GREEN after.\n- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router)\n- Live: n/a: unit/fixture-level verification; no live deployment was run", "tags": [ "controller", "apiserver", @@ -18500,16 +18504,20 @@ "related_bugs": [ "bug-1938" ], - "fix_commits": [], + "fix_commits": [ + "b239db5f", + "1ff068f6", + "4f579041" + ], "record": ".wolf/bugs/bug-1960.md", "gitea_issue": 391, "gitea_synced_status": "open" }, { "id": "bug-1961", - "status": "open", + "status": "fixed", "status_updated": "2026-10-10", - "status_note": "newly logged", + "status_note": "", "component": "apiserver", "severity": "wrong-result", "timestamp": "2026-10-10T10:16:04Z", @@ -18517,23 +18525,25 @@ "occurrences": 1, "error_message": "apiserver ControlHost writes a second response after validateHost rejected the request\n\n**Observed:** (source reading; follow-up of the 2026-10-10 P0 fixes (bug-1936..1957))\nReported by the bug-1949 fix: `validateHost` (`controllers/hosts.go`) writes a 403 via `ReturnError` without aborting and `ControlHost` continues and writes another response (today masked by the route's RequirePermission).\n\n**Expected:** A rejected host control stops after the first error response.\n**Impact:** Wrong result / noise: double responses, possible action after a rejection if the route guard changes.\n**Found by:** follow-up of the 2026-10-10 P0 fixes (bug-1936..1957)", "file": "src/frontend/apiserver/controllers/hosts.go", - "root_cause": "TODO: required for fixed/partial", - "fix": "TODO: required for fixed/partial\n\nVerification:\n- Test: TODO: name + file; RED before the fix, GREEN after\n- Suites: TODO: suite results and build profile\n- Live: TODO: soak/deploy run, or n/a and why", + "root_cause": "`validateHost` (`controllers/hosts.go`) wrote a 403 via `ReturnError` without aborting, so `ControlHost`'s `IsAborted()` check never fired and it wrote a second (202) response. Pre-existing.", + "fix": "`validateHost` no longer writes a response: it checks `host:control` once, before the host loop (it used to check per host and not at all for an empty list), and returns an error; `ControlHost` writes a single 403 and returns; for `all` the host list is resolved first.\n\nVerification:\n- Test: `TestControlHost_RejectedWritesSingleResponse` (`controllers/hosts_handler_test.go`, route without its guard): RED `hosts_handler_test.go:61: handler wrote 2 responses, want 1`, GREEN after.\n- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router)\n- Live: n/a: unit/fixture-level verification; no live deployment was run", "tags": [ "apiserver", "hosts" ], "related_bugs": [], - "fix_commits": [], + "fix_commits": [ + "d720132c" + ], "record": ".wolf/bugs/bug-1961.md", "gitea_issue": 392, "gitea_synced_status": "open" }, { "id": "bug-1962", - "status": "open", + "status": "fixed", "status_updated": "2026-10-10", - "status_note": "newly logged", + "status_note": "", "component": "apiserver", "severity": "security", "timestamp": "2026-10-10T10:16:07Z", @@ -18541,24 +18551,26 @@ "occurrences": 1, "error_message": "apiserver JWT excluded-path check drops the last character of every entry, matching unintended paths\n\n**Observed:** (source reading; follow-up of the 2026-10-10 P0 fixes (bug-1936..1957))\n`ValidateHandler.isExcludedPath` (`middleware/validatehandler.go:52-61`) trims the last character of every excluded entry, not only a trailing `*`, and matches by prefix: `/ping` matches any path starting with `/pin`, and `/swagger/*any` never matches.\n\n**Expected:** Exact entries match exactly; only entries ending in `*` match by prefix.\n**Impact:** Security: paths that merely share a prefix with an excluded entry skip JWT validation.\n**Found by:** follow-up of the 2026-10-10 P0 fixes (bug-1936..1957)", "file": "src/frontend/apiserver/middleware/validatehandler.go", - "root_cause": "TODO: required for fixed/partial", - "fix": "TODO: required for fixed/partial\n\nVerification:\n- Test: TODO: name + file; RED before the fix, GREEN after\n- Suites: TODO: suite results and build profile\n- Live: TODO: soak/deploy run, or n/a and why", + "root_cause": "`isExcludedPath` (`middleware/validatehandler.go`) cut the last character of every excluded entry and matched by prefix. Pre-existing.", + "fix": "The (hardcoded) excluded list moved to `jwtExcludedPaths`; entries match exactly; an entry ending in a gin catch-all segment (`/swagger/*any`) matches anything under `/swagger/`; ping, logon, logout and swagger still excluded.\n\nVerification:\n- Test: `TestIsExcludedPath`, `TestValidateHandler_PrefixLookAlikeNeedsToken` (`middleware/validatehandler_test.go`): RED `isExcludedPath(\"/pinX\") = true, want false`, `isExcludedPath(\"/swagger/index.html\") = false, want true`, `/pinX without a token reached the handler (status 200)`; GREEN after.\n- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router)\n- Live: n/a: unit/fixture-level verification; no live deployment was run", "tags": [ "apiserver", "security", "jwt" ], "related_bugs": [], - "fix_commits": [], + "fix_commits": [ + "387a421c" + ], "record": ".wolf/bugs/bug-1962.md", "gitea_issue": 393, "gitea_synced_status": "open" }, { "id": "bug-1963", - "status": "investigating", + "status": "fixed", "status_updated": "2026-10-10", - "status_note": "suspected; confirm with a scheduler fixture test", + "status_note": "", "component": "scheduler", "severity": "wrong-result", "timestamp": "2026-10-10T10:16:09Z", @@ -18566,23 +18578,25 @@ "occurrences": 1, "error_message": "xce_scheduler host iteration offers unavailable or reconciling hosts as placement candidates\n\n**Observed:** (source reading; follow-up of the 2026-10-10 P0 fixes (bug-1936..1957))\nSuspected (reported by the bug-1939 fix, not verified): `sched_host_iter_next()` (`src/scheduler/src/host_iter.c`) skips admin-closed hosts since bug-1939 but not `XCE_WORKER_UNAVAIL` or `XCE_WORKER_RECONCILING`; other checks (`resource.c:704,887`, `core_session.c:136`) are accounting skips, not placement filters.\n\n**Expected:** Unavailable and reconciling hosts get no new placements.\n**Impact:** Possible dispatch to a dead or reconciling host (dispatch then fails and is retried).\n**Found by:** follow-up of the 2026-10-10 P0 fixes (bug-1936..1957)", "file": "src/scheduler/src/host_iter.c", - "root_cause": "TODO: required for fixed/partial", - "fix": "TODO: required for fixed/partial\n\nVerification:\n- Test: TODO: name + file; RED before the fix, GREEN after\n- Suites: TODO: suite results and build profile\n- Live: TODO: soak/deploy run, or n/a and why", + "root_cause": "`sched_host_iter_next()` (`src/scheduler/src/host_iter.c`) skipped only admin-closed hosts; the UNAVAIL/RECONCILING checks elsewhere (`resource.c`, `core_session.c`) only affect accounting, and such a host keeps its last usage, so the resource check does not reject it. Pre-existing.", + "fix": "`host_iter.c` skips `IS_WORKER_UNAVAIL` and `IS_WORKER_RECONCILING` hosts right after the closed-host check, for both drivers.\n\nVerification:\n- Test: `test_unavail_and_reconciling_hosts_get_no_new_jobs` (`src/scheduler/testsuite/test_host_unavail_reconciling.c`): RED `test_host_unavail_reconciling.c:200: FAIL: Expected 1 Was 2 (normal mode, unavailable host: serial job B must pend)`, GREEN after (both drivers, both status bits).\n- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router)\n- Live: n/a: unit/fixture-level verification; no live deployment was run", "tags": [ "scheduler", "host" ], "related_bugs": [], - "fix_commits": [], + "fix_commits": [ + "2cd02cdc" + ], "record": ".wolf/bugs/bug-1963.md", "gitea_issue": 394, "gitea_synced_status": "investigating" }, { "id": "bug-1964", - "status": "open", + "status": "fixed", "status_updated": "2026-10-10", - "status_note": "newly logged", + "status_note": "", "component": "scheduler", "severity": "wrong-result", "timestamp": "2026-10-10T10:16:12Z", @@ -18590,8 +18604,8 @@ "occurrences": 1, "error_message": "xce_scheduler --sim does not re-seed limits when a config push arrives mid-replay\n\n**Observed:** (source reading; follow-up of the 2026-10-10 P0 fixes (bug-1936..1957))\nReported by the bug-1943 fix: the simulator seeds the limit plugin only at bootstrap (`sim_bootstrap_limit_plugin`); a CONFIG_PUSH during replay updates groups (bug-1943) but a new or changed limit config never reaches the limit plugin, while the live scheduler applies it (`plugin_config_update`).\n\n**Expected:** Live and simulated scheduling apply the same limits after a config change.\n**Impact:** Replays that span a limit config change decide differently than live.\n**Found by:** follow-up of the 2026-10-10 P0 fixes (bug-1936..1957)", "file": "src/scheduler/src/sim_limit.c, src/scheduler/src/core_events.c", - "root_cause": "TODO: required for fixed/partial", - "fix": "TODO: required for fixed/partial\n\nVerification:\n- Test: TODO: name + file; RED before the fix, GREEN after\n- Suites: TODO: suite results and build profile\n- Live: TODO: soak/deploy run, or n/a and why", + "root_cause": "Live `sched_init_config()` calls `plugin_config_update()`, but the core's CONFIG_PUSH handler (`core_events.c:xce_core_push_events`) did not, and the simulator loaded limits only at bootstrap (`sim_limit.c:sim_bootstrap_limit_plugin`). Pre-existing.", + "fix": "`plugin_config_update_core(core, config)` (`plugin.c/.h`) is the shared update (live's `plugin_config_update()` calls it); a driver-registered core hook (`xce_core_set_config_cb`, `core.h`/`core_session.c`, modelled on the cache-clear hook, keeping the core free of plugin calls) is called by CONFIG_PUSH after loading groups; the simulator registers it; `limit_update_config()` reads `plugin->core` instead of casting `plugin->sched` (absent in sim); fixtures set `plugin->core`.\n\nVerification:\n- Test: `test_config_push_reaches_limit_plugin_in_sim_as_in_live` (`src/scheduler/testsuite/test_sim_limit_config_push_parity.c`): RED `test_sim_limit_config_push_parity.c:311: FAIL: Expected 1 Was 0 (sim: a CONFIG_PUSH during replay must configure its limit as live does)`, GREEN after; `test_sim_group_limit_parity`, `test_limit_plugin` (6/6), `test_repl_limit_table` (35/35) pass.\n- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router)\n- Live: n/a: unit/fixture-level verification; no live deployment was run", "tags": [ "scheduler", "sim", @@ -18601,16 +18615,18 @@ "related_bugs": [ "bug-1943" ], - "fix_commits": [], + "fix_commits": [ + "3224e906" + ], "record": ".wolf/bugs/bug-1964.md", "gitea_issue": 395, "gitea_synced_status": "open" }, { "id": "bug-1965", - "status": "open", + "status": "fixed", "status_updated": "2026-10-10", - "status_note": "newly logged", + "status_note": "", "component": "build", "severity": "noise", "timestamp": "2026-10-10T10:16:14Z", @@ -18618,24 +18634,26 @@ "occurrences": 1, "error_message": "xce-build.sh apiserver skips the middleware, router and cmd Go test packages\n\n**Observed:** (source reading; follow-up of the 2026-10-10 P0 fixes (bug-1936..1957))\n`xce-build.sh` runs `go test ./store/... ./utils/... ./controllers/... ./backend/...` for the apiserver; `./middleware/...`, `./router/...`, `./cmd/...` (which hold several security tests since bug-1945..1952) are never run by the build.\n\n**Expected:** The build runs every apiserver test package.\n**Impact:** Regressions in middleware/router/cmd tests go unnoticed by `/xce-build`.\n**Found by:** follow-up of the 2026-10-10 P0 fixes (bug-1936..1957)", "file": "src/tools/xce-build.sh", - "root_cause": "TODO: required for fixed/partial", - "fix": "TODO: required for fixed/partial\n\nVerification:\n- Test: TODO: name + file; RED before the fix, GREEN after\n- Suites: TODO: suite results and build profile\n- Live: TODO: soak/deploy run, or n/a and why", + "root_cause": "`src/tools/xce-build.sh` listed `./store/... ./utils/... ./controllers/... ./backend/...` for the apiserver Go tests, so `cmd`, `middleware`, `models` and `router` never ran in the build. Pre-existing.", + "fix": "Both apiserver `go test` lines (components `apiserver` and `all`) now run `./...` (vendor excluded automatically).\n\nVerification:\n- Test: With a temporary deliberately failing middleware test (not committed): the old script exited 0 without running middleware; the new script exited 1 with `--- FAIL: TestBug1965DeliberatelyBroken` / `FAIL apiserver/middleware`. The merged-branch `all` build log shows `ok apiserver/cmd`, `middleware`, `models`, `router`.\n- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router)\n- Live: n/a: unit/fixture-level verification; no live deployment was run", "tags": [ "build", "tests", "apiserver" ], "related_bugs": [], - "fix_commits": [], + "fix_commits": [ + "387cc994" + ], "record": ".wolf/bugs/bug-1965.md", "gitea_issue": 396, "gitea_synced_status": "open" }, { "id": "bug-1966", - "status": "open", + "status": "fixed", "status_updated": "2026-10-10", - "status_note": "newly logged", + "status_note": "", "component": "apiserver", "severity": "noise", "timestamp": "2026-10-10T10:16:16Z", @@ -18643,18 +18661,46 @@ "occurrences": 1, "error_message": "apiserver swagger generation fails on hosts.go models.Error; published swagger is a stale stub\n\n**Observed:** (source reading; follow-up of the 2026-10-10 P0 fixes (bug-1936..1957))\n`swag init -g cmd/main.go` fails: `ParseComment error in controllers/hosts.go for comment: '// @Failure 400 {object} models.Error': cannot find type definition: models.Error`. `docs/swagger.yaml` is a 212-line stub last generated before the meson merge (no job-control endpoints).\n\n**Expected:** Swagger generates from the annotations and the published spec covers the API.\n**Impact:** API documentation missing/stale.\n**Found by:** follow-up of the 2026-10-10 P0 fixes (bug-1936..1957)", "file": "src/frontend/apiserver/controllers/hosts.go, src/frontend/apiserver/docs/swagger.yaml", - "root_cause": "TODO: required for fixed/partial", - "fix": "TODO: required for fixed/partial\n\nVerification:\n- Test: TODO: name + file; RED before the fix, GREEN after\n- Suites: TODO: suite results and build profile\n- Live: TODO: soak/deploy run, or n/a and why", + "root_cause": "Swag annotations in `controllers/hosts.go` and `controllers/jobs.go` named non-existent types (`models.Error`, `ErrorResponse`, `JobsResponse`, `JobSignalRequest`, `JobControlResponseSummary`), had malformed `@Param`/`@Failure` lines and wrong `/api/v1` paths, and no general API info existed. Pre-existing.", + "fix": "Errors use `JsonErrStruct` (what `ReturnError` writes), success `JsonStruct`; bodies use `models.JobSubmitRequest`, `JobControlJobIDRequest`, `JobControlRequest`; paths match the routes; `@BasePath /xce/v1` and general info in `cmd/main.go`; `docs/docs.go`, `swagger.json`, `swagger.yaml` regenerated with `swag init -g cmd/main.go -o docs`. Only hosts and jobs handlers are annotated so far.\n\nVerification:\n- Test: RED: `swag init -g cmd/main.go` failed with `ParseComment error in file .../controllers/hosts.go ... models.Error: cannot find type definition`; GREEN: swag runs cleanly and the spec lists `/hosts`, `/jobs`, `/jobs/create`, `/jobs/{jobids}`, `/jobs/{jobids}/control`, `/jobs/control`; `cmd` (imports docs) tests pass.\n- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router)\n- Live: n/a: unit/fixture-level verification; no live deployment was run", "tags": [ "apiserver", "docs", "swagger" ], "related_bugs": [], - "fix_commits": [], + "fix_commits": [ + "f7fef278" + ], "record": ".wolf/bugs/bug-1966.md", "gitea_issue": 397, "gitea_synced_status": "open" + }, + { + "id": "bug-1967", + "status": "open", + "status_updated": "2026-10-10", + "status_note": "newly logged", + "component": "scheduler", + "severity": "wrong-result", + "timestamp": "2026-10-10T10:47:20Z", + "last_seen": "2026-10-10T10:47:20Z", + "occurrences": 1, + "error_message": "xce_scheduler ignores limit changes in a re-pushed config: seeding stops at the first existing name\n\n**Observed:** (source reading, 2026-10-10; reported by the bug-1964 fix)\n`limit_seed_config_impl()` (`src/scheduler/src/plugins/limit.c`) builds each\nconfigured limit and adds it with `key_add(privdata->limits, limit->name, ...)`,\nreturning at the first failure:\n```c\nrc = key_add(privdata->limits, (void*)limit->name, (void*)limit);\n...\nif (rc < 0) {\n ... /* bug-1685: the kvmgr already freed `limit` */\n return rc;\n}\n```\nOn a later config push (live config upgrade, or a mid-replay CONFIG_PUSH in\nthe simulator since bug-1964) every limit name already exists, so the first\n`key_add` fails with `-EEXIST`: no limit is updated, a limit whose value or\nscope changed keeps its old definition, and limits listed after an existing\none are never added.\n\n**Expected:** a config push replaces the configured limits: changed limits\ntake their new definition, new ones are added, removed ones disappear\n(their usage instances handled consistently).\n**Impact:** wrong result: limit changes made by a live config upgrade do not\ntake effect until the scheduler restarts; live and simulator agree (both\nwrong).\n**Found by:** fix of bug-1964 (sim limit config push), 2026-10-10.", + "file": "src/scheduler/src/plugins/limit.c", + "root_cause": "TODO: required for fixed/partial", + "fix": "TODO: required for fixed/partial\n\nVerification:\n- Test: TODO: name + file; RED before the fix, GREEN after\n- Suites: TODO: suite results and build profile\n- Live: TODO: soak/deploy run, or n/a and why", + "tags": [ + "scheduler", + "limit", + "config" + ], + "related_bugs": [ + "bug-1964", + "bug-1685" + ], + "fix_commits": [], + "record": ".wolf/bugs/bug-1967.md" } ] } diff --git a/.wolf/bugs/bug-1958.md b/.wolf/bugs/bug-1958.md index 3444d1e8..5b677893 100644 --- a/.wolf/bugs/bug-1958.md +++ b/.wolf/bugs/bug-1958.md @@ -1,8 +1,8 @@ --- id: "bug-1958" -status: "open" +status: "fixed" status_updated: "2026-10-10" -status_note: "newly logged" +status_note: "" component: "apiserver" severity: "security" timestamp: "2026-10-10T10:15:56Z" @@ -11,6 +11,7 @@ occurrences: 1 files: ["src/frontend/apiserver/controllers/roles.go"] tags: ["apiserver", "security", "rbac"] related: ["bug-1952"] +fix_commits: ["736ed81d"] gitea_issue: 389 gitea_synced_status: "open" gitea_synced_hash: "499c8864257a" @@ -39,15 +40,16 @@ Reported by the bug-1952 fix: role create/update in `controllers/roles.go` accep - Frequency: deterministic (by code reading; confirmed by the RED test of the fix) - Evidence: the code sites quoted in Description; no logs kept. ## Root cause -TODO: required for fixed/partial +`CreateRole`, `AddPermissionsForRole` and `DeleteRoleAndPermissions` (`controllers/roles.go`) were guarded only by the route's `RequirePermission("rbac:control")`, which accepts any scope, so a `self`-scoped grant could add `cluster:control` to a role its holder has. Pre-existing. ## Fix -TODO: required for fixed/partial +New `checkRoleChangePermission()` in `roles.go`, called first in the three handlers, reuses `canChangeUserRoles` (bug-1952; all grants before deduplication) and answers 403 unless the caller holds `rbac:control` with scope `all`. ## Verification -- Test: TODO: name + file; RED before the fix, GREEN after -- Suites: TODO: suite results and build profile -- Live: TODO: soak/deploy run, or n/a and why +- Test: `TestAddPermissionsForRole_SelfScopeCannotExtendOwnRole` (`controllers/roles_rbac_test.go`): RED `roles_rbac_test.go:65: privilege escalation: mallory (rbac:control scope self) added cluster:control to its own role (status 200)`; `TestCreateRole_SelfScopeRejected`, `TestDeleteRole_SelfScopeRejected` RED (status 200, want 403); GREEN after, plus `TestRoleChanges_AllScopeAllowed`. +- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router) +- Live: n/a: unit/fixture-level verification; no live deployment was run ## References -- Related: none +- Fix commits: `736ed81d` (on `scheduler-framework`) +- Found as a follow-up of the 2026-10-10 P0 fixes (bug-1936..1957) diff --git a/.wolf/bugs/bug-1959.md b/.wolf/bugs/bug-1959.md index c36e56c8..6638ce7f 100644 --- a/.wolf/bugs/bug-1959.md +++ b/.wolf/bugs/bug-1959.md @@ -1,8 +1,8 @@ --- id: "bug-1959" -status: "open" +status: "fixed" status_updated: "2026-10-10" -status_note: "newly logged" +status_note: "" component: "xcecli" severity: "security" timestamp: "2026-10-10T10:15:59Z" @@ -11,6 +11,7 @@ occurrences: 1 files: ["src/frontend/xcecli/cmd/auth"] tags: ["xcecli", "security", "jwt"] related: ["bug-1946"] +fix_commits: ["03798230"] gitea_issue: 390 gitea_synced_status: "open" gitea_synced_hash: "3ee2d40b62c8" @@ -39,15 +40,16 @@ Reported by the bug-1946 fix: server-side token revocation exists (`POST /xce/v1 - Frequency: deterministic (by code reading; confirmed by the RED test of the fix) - Evidence: the code sites quoted in Description; no logs kept. ## Root cause -TODO: required for fixed/partial +`runLogout` (`src/frontend/xcecli/cmd/auth/logout.go`) only removed the server entry from the local config; the CLI never called the logout API, so the token stayed valid server side until expiry. Pre-existing. ## Fix -TODO: required for fixed/partial +New `APIClient.Logout(token)` (`common/request/request.go`) posts `/xce/v1/auth/logout` with the bearer token; `runLogout` calls it first and, if the server is unreachable or rejects the token, warns `server-side token revocation could not be confirmed` and still clears local state; help text updated. ## Verification -- Test: TODO: name + file; RED before the fix, GREEN after -- Suites: TODO: suite results and build profile -- Live: TODO: soak/deploy run, or n/a and why +- Test: `TestRunLogout_RevokesTokenOnServer` (`cmd/auth/logout_test.go`, httptest): RED `logout_test.go:124: requests = [], want ["POST /xce/v1/auth/logout Bearer tok"]`; `TestRunLogout_ServerFailureStillClearsLocalState` (rejected and unreachable server) RED `logout_test.go:157`; GREEN after (5 auth tests). +- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router) +- Live: n/a: unit/fixture-level verification; no live deployment was run ## References -- Related: none +- Fix commits: `03798230` (on `scheduler-framework`) +- Found as a follow-up of the 2026-10-10 P0 fixes (bug-1936..1957) diff --git a/.wolf/bugs/bug-1960.md b/.wolf/bugs/bug-1960.md index 749a9346..8542ddd0 100644 --- a/.wolf/bugs/bug-1960.md +++ b/.wolf/bugs/bug-1960.md @@ -1,8 +1,8 @@ --- id: "bug-1960" -status: "open" +status: "fixed" status_updated: "2026-10-10" -status_note: "newly logged" +status_note: "" component: "controller" severity: "wrong-result" timestamp: "2026-10-10T10:16:02Z" @@ -11,6 +11,7 @@ occurrences: 1 files: ["src/controller/job.c", "src/frontend/apiserver/controllers/jobs.go"] tags: ["controller", "apiserver", "queue", "admin-control"] related: ["bug-1938"] +fix_commits: ["b239db5f", "1ff068f6", "4f579041"] gitea_issue: 391 gitea_synced_status: "open" gitea_synced_hash: "e86054beb851" @@ -39,15 +40,16 @@ Reported by the bug-1938 fix: `handle_jobsubmit()` (`src/controller/job.c`) neve - Frequency: deterministic (by code reading; confirmed by the RED test of the fix) - Evidence: the code sites quoted in Description; no logs kept. ## Root cause -TODO: required for fixed/partial +`handle_jobsubmit()` (`src/controller/job.c`) only checked that the queue field was non-empty, never the queue's existence or `QUEUE_STAT_CLOSED`, so single and array submissions from libxce/harness (and REST without a queue name) reached closed or unknown queues; `preCheckJobRequest()` (`controllers/jobs.go`) checked only a named queue. Pre-existing. ## Fix -TODO: required for fixed/partial +Controller (`b239db5f`): new `jobsubmit_check_queue()` (in-memory read under the qmgr/queue read locks) rejects an unknown queue or `lost_and_found` with `-E_XCE_BAD_QUEUE` and a closed queue with new `-E_XCE_QUEUE_CLOSED` (10016, `include/xce/errnox.h`, string in `lib/utils.c`) before the hook, the in-flight registry and any WAL append; replay/apply paths unchanged, so committed submits still apply. Apiserver (`1ff068f6`): the default queue gets the same existence/closed checks. Apiserver (`4f579041`): `SubmitJob` maps the two codes to `backend.ErrQueueClosed`/`ErrQueueInvalid` (checked against errnox.h at compile time) and the submit/restart handlers answer 400 naming the queue (covers a close racing the apiserver check). Behaviour change: a submission without a queue is rejected when the cluster has no `normal` queue. ## Verification -- Test: TODO: name + file; RED before the fix, GREEN after -- Suites: TODO: suite results and build profile -- Live: TODO: soak/deploy run, or n/a and why +- Test: Controller: `test_jobsubmit_closed_queue_rejected`, `test_jobsubmit_unknown_queue_rejected`, `test_jobsubmit_lost_and_found_rejected` (`src/controller/testsuite/test_jobsubmit_validate.c`): RED `test_jobsubmit_validate.c:370: FAIL: Expected -10016 Was -5` (and -10008), GREEN after; `test_ace_job_submit_closed_queue_still_applies` guards replay. Apiserver: `jobs_test.go:137: submission to the closed default queue: err = ` RED, GREEN after; `TestSubmitRCError`, `TestJobHandler_SubmitQueueRejectedByController` failed to build before (no typed errors), GREEN after. +- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router) +- Live: n/a: unit/fixture-level verification; no live deployment was run ## References -- Related: none +- Fix commits: `b239db5f`, `1ff068f6`, `4f579041` (on `scheduler-framework`) +- Found as a follow-up of the 2026-10-10 P0 fixes (bug-1936..1957) diff --git a/.wolf/bugs/bug-1961.md b/.wolf/bugs/bug-1961.md index 73d37fc8..e5b225ff 100644 --- a/.wolf/bugs/bug-1961.md +++ b/.wolf/bugs/bug-1961.md @@ -1,8 +1,8 @@ --- id: "bug-1961" -status: "open" +status: "fixed" status_updated: "2026-10-10" -status_note: "newly logged" +status_note: "" component: "apiserver" severity: "wrong-result" timestamp: "2026-10-10T10:16:04Z" @@ -11,6 +11,7 @@ occurrences: 1 files: ["src/frontend/apiserver/controllers/hosts.go"] tags: ["apiserver", "hosts"] related: [] +fix_commits: ["d720132c"] gitea_issue: 392 gitea_synced_status: "open" gitea_synced_hash: "7371293540aa" @@ -38,15 +39,16 @@ Reported by the bug-1949 fix: `validateHost` (`controllers/hosts.go`) writes a 4 - Frequency: deterministic (by code reading; confirmed by the RED test of the fix) - Evidence: the code sites quoted in Description; no logs kept. ## Root cause -TODO: required for fixed/partial +`validateHost` (`controllers/hosts.go`) wrote a 403 via `ReturnError` without aborting, so `ControlHost`'s `IsAborted()` check never fired and it wrote a second (202) response. Pre-existing. ## Fix -TODO: required for fixed/partial +`validateHost` no longer writes a response: it checks `host:control` once, before the host loop (it used to check per host and not at all for an empty list), and returns an error; `ControlHost` writes a single 403 and returns; for `all` the host list is resolved first. ## Verification -- Test: TODO: name + file; RED before the fix, GREEN after -- Suites: TODO: suite results and build profile -- Live: TODO: soak/deploy run, or n/a and why +- Test: `TestControlHost_RejectedWritesSingleResponse` (`controllers/hosts_handler_test.go`, route without its guard): RED `hosts_handler_test.go:61: handler wrote 2 responses, want 1`, GREEN after. +- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router) +- Live: n/a: unit/fixture-level verification; no live deployment was run ## References -- Related: none +- Fix commits: `d720132c` (on `scheduler-framework`) +- Found as a follow-up of the 2026-10-10 P0 fixes (bug-1936..1957) diff --git a/.wolf/bugs/bug-1962.md b/.wolf/bugs/bug-1962.md index 3eab3a3e..2e54dfdd 100644 --- a/.wolf/bugs/bug-1962.md +++ b/.wolf/bugs/bug-1962.md @@ -1,8 +1,8 @@ --- id: "bug-1962" -status: "open" +status: "fixed" status_updated: "2026-10-10" -status_note: "newly logged" +status_note: "" component: "apiserver" severity: "security" timestamp: "2026-10-10T10:16:07Z" @@ -11,6 +11,7 @@ occurrences: 1 files: ["src/frontend/apiserver/middleware/validatehandler.go"] tags: ["apiserver", "security", "jwt"] related: [] +fix_commits: ["387a421c"] gitea_issue: 393 gitea_synced_status: "open" gitea_synced_hash: "8d6ac245ba9a" @@ -38,15 +39,16 @@ gitea_synced_hash: "8d6ac245ba9a" - Frequency: deterministic (by code reading; confirmed by the RED test of the fix) - Evidence: the code sites quoted in Description; no logs kept. ## Root cause -TODO: required for fixed/partial +`isExcludedPath` (`middleware/validatehandler.go`) cut the last character of every excluded entry and matched by prefix. Pre-existing. ## Fix -TODO: required for fixed/partial +The (hardcoded) excluded list moved to `jwtExcludedPaths`; entries match exactly; an entry ending in a gin catch-all segment (`/swagger/*any`) matches anything under `/swagger/`; ping, logon, logout and swagger still excluded. ## Verification -- Test: TODO: name + file; RED before the fix, GREEN after -- Suites: TODO: suite results and build profile -- Live: TODO: soak/deploy run, or n/a and why +- Test: `TestIsExcludedPath`, `TestValidateHandler_PrefixLookAlikeNeedsToken` (`middleware/validatehandler_test.go`): RED `isExcludedPath("/pinX") = true, want false`, `isExcludedPath("/swagger/index.html") = false, want true`, `/pinX without a token reached the handler (status 200)`; GREEN after. +- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router) +- Live: n/a: unit/fixture-level verification; no live deployment was run ## References -- Related: none +- Fix commits: `387a421c` (on `scheduler-framework`) +- Found as a follow-up of the 2026-10-10 P0 fixes (bug-1936..1957) diff --git a/.wolf/bugs/bug-1963.md b/.wolf/bugs/bug-1963.md index 7825f41e..92ac64f0 100644 --- a/.wolf/bugs/bug-1963.md +++ b/.wolf/bugs/bug-1963.md @@ -1,8 +1,8 @@ --- id: "bug-1963" -status: "investigating" +status: "fixed" status_updated: "2026-10-10" -status_note: "suspected; confirm with a scheduler fixture test" +status_note: "" component: "scheduler" severity: "wrong-result" timestamp: "2026-10-10T10:16:09Z" @@ -11,6 +11,7 @@ occurrences: 1 files: ["src/scheduler/src/host_iter.c"] tags: ["scheduler", "host"] related: [] +fix_commits: ["2cd02cdc"] gitea_issue: 394 gitea_synced_status: "investigating" gitea_synced_hash: "6f389c3e4c8b" @@ -38,15 +39,16 @@ Suspected (reported by the bug-1939 fix, not verified): `sched_host_iter_next()` - Frequency: deterministic (by code reading; confirmed by the RED test of the fix) - Evidence: the code sites quoted in Description; no logs kept. ## Root cause -TODO: required for fixed/partial +`sched_host_iter_next()` (`src/scheduler/src/host_iter.c`) skipped only admin-closed hosts; the UNAVAIL/RECONCILING checks elsewhere (`resource.c`, `core_session.c`) only affect accounting, and such a host keeps its last usage, so the resource check does not reject it. Pre-existing. ## Fix -TODO: required for fixed/partial +`host_iter.c` skips `IS_WORKER_UNAVAIL` and `IS_WORKER_RECONCILING` hosts right after the closed-host check, for both drivers. ## Verification -- Test: TODO: name + file; RED before the fix, GREEN after -- Suites: TODO: suite results and build profile -- Live: TODO: soak/deploy run, or n/a and why +- Test: `test_unavail_and_reconciling_hosts_get_no_new_jobs` (`src/scheduler/testsuite/test_host_unavail_reconciling.c`): RED `test_host_unavail_reconciling.c:200: FAIL: Expected 1 Was 2 (normal mode, unavailable host: serial job B must pend)`, GREEN after (both drivers, both status bits). +- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router) +- Live: n/a: unit/fixture-level verification; no live deployment was run ## References -- Related: none +- Fix commits: `2cd02cdc` (on `scheduler-framework`) +- Found as a follow-up of the 2026-10-10 P0 fixes (bug-1936..1957) diff --git a/.wolf/bugs/bug-1964.md b/.wolf/bugs/bug-1964.md index 13fe8a55..c334eab6 100644 --- a/.wolf/bugs/bug-1964.md +++ b/.wolf/bugs/bug-1964.md @@ -1,8 +1,8 @@ --- id: "bug-1964" -status: "open" +status: "fixed" status_updated: "2026-10-10" -status_note: "newly logged" +status_note: "" component: "scheduler" severity: "wrong-result" timestamp: "2026-10-10T10:16:12Z" @@ -11,6 +11,7 @@ occurrences: 1 files: ["src/scheduler/src/sim_limit.c", "src/scheduler/src/core_events.c"] tags: ["scheduler", "sim", "equivalence", "limit"] related: ["bug-1943"] +fix_commits: ["3224e906"] gitea_issue: 395 gitea_synced_status: "open" gitea_synced_hash: "40588533d766" @@ -38,15 +39,16 @@ Reported by the bug-1943 fix: the simulator seeds the limit plugin only at boots - Frequency: deterministic (by code reading; confirmed by the RED test of the fix) - Evidence: the code sites quoted in Description; no logs kept. ## Root cause -TODO: required for fixed/partial +Live `sched_init_config()` calls `plugin_config_update()`, but the core's CONFIG_PUSH handler (`core_events.c:xce_core_push_events`) did not, and the simulator loaded limits only at bootstrap (`sim_limit.c:sim_bootstrap_limit_plugin`). Pre-existing. ## Fix -TODO: required for fixed/partial +`plugin_config_update_core(core, config)` (`plugin.c/.h`) is the shared update (live's `plugin_config_update()` calls it); a driver-registered core hook (`xce_core_set_config_cb`, `core.h`/`core_session.c`, modelled on the cache-clear hook, keeping the core free of plugin calls) is called by CONFIG_PUSH after loading groups; the simulator registers it; `limit_update_config()` reads `plugin->core` instead of casting `plugin->sched` (absent in sim); fixtures set `plugin->core`. ## Verification -- Test: TODO: name + file; RED before the fix, GREEN after -- Suites: TODO: suite results and build profile -- Live: TODO: soak/deploy run, or n/a and why +- Test: `test_config_push_reaches_limit_plugin_in_sim_as_in_live` (`src/scheduler/testsuite/test_sim_limit_config_push_parity.c`): RED `test_sim_limit_config_push_parity.c:311: FAIL: Expected 1 Was 0 (sim: a CONFIG_PUSH during replay must configure its limit as live does)`, GREEN after; `test_sim_group_limit_parity`, `test_limit_plugin` (6/6), `test_repl_limit_table` (35/35) pass. +- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router) +- Live: n/a: unit/fixture-level verification; no live deployment was run ## References -- Related: none +- Fix commits: `3224e906` (on `scheduler-framework`) +- Found as a follow-up of the 2026-10-10 P0 fixes (bug-1936..1957) diff --git a/.wolf/bugs/bug-1965.md b/.wolf/bugs/bug-1965.md index 55394bd8..445a54b2 100644 --- a/.wolf/bugs/bug-1965.md +++ b/.wolf/bugs/bug-1965.md @@ -1,8 +1,8 @@ --- id: "bug-1965" -status: "open" +status: "fixed" status_updated: "2026-10-10" -status_note: "newly logged" +status_note: "" component: "build" severity: "noise" timestamp: "2026-10-10T10:16:14Z" @@ -11,6 +11,7 @@ occurrences: 1 files: ["src/tools/xce-build.sh"] tags: ["build", "tests", "apiserver"] related: [] +fix_commits: ["387cc994"] gitea_issue: 396 gitea_synced_status: "open" gitea_synced_hash: "eed0237b5ae0" @@ -38,15 +39,16 @@ gitea_synced_hash: "eed0237b5ae0" - Frequency: deterministic (by code reading; confirmed by the RED test of the fix) - Evidence: the code sites quoted in Description; no logs kept. ## Root cause -TODO: required for fixed/partial +`src/tools/xce-build.sh` listed `./store/... ./utils/... ./controllers/... ./backend/...` for the apiserver Go tests, so `cmd`, `middleware`, `models` and `router` never ran in the build. Pre-existing. ## Fix -TODO: required for fixed/partial +Both apiserver `go test` lines (components `apiserver` and `all`) now run `./...` (vendor excluded automatically). ## Verification -- Test: TODO: name + file; RED before the fix, GREEN after -- Suites: TODO: suite results and build profile -- Live: TODO: soak/deploy run, or n/a and why +- Test: With a temporary deliberately failing middleware test (not committed): the old script exited 0 without running middleware; the new script exited 1 with `--- FAIL: TestBug1965DeliberatelyBroken` / `FAIL apiserver/middleware`. The merged-branch `all` build log shows `ok apiserver/cmd`, `middleware`, `models`, `router`. +- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router) +- Live: n/a: unit/fixture-level verification; no live deployment was run ## References -- Related: none +- Fix commits: `387cc994` (on `scheduler-framework`) +- Found as a follow-up of the 2026-10-10 P0 fixes (bug-1936..1957) diff --git a/.wolf/bugs/bug-1966.md b/.wolf/bugs/bug-1966.md index 482e8fce..61f52e45 100644 --- a/.wolf/bugs/bug-1966.md +++ b/.wolf/bugs/bug-1966.md @@ -1,8 +1,8 @@ --- id: "bug-1966" -status: "open" +status: "fixed" status_updated: "2026-10-10" -status_note: "newly logged" +status_note: "" component: "apiserver" severity: "noise" timestamp: "2026-10-10T10:16:16Z" @@ -11,6 +11,7 @@ occurrences: 1 files: ["src/frontend/apiserver/controllers/hosts.go", "src/frontend/apiserver/docs/swagger.yaml"] tags: ["apiserver", "docs", "swagger"] related: [] +fix_commits: ["f7fef278"] gitea_issue: 397 gitea_synced_status: "open" gitea_synced_hash: "d98430bca460" @@ -38,15 +39,16 @@ gitea_synced_hash: "d98430bca460" - Frequency: deterministic (by code reading; confirmed by the RED test of the fix) - Evidence: the code sites quoted in Description; no logs kept. ## Root cause -TODO: required for fixed/partial +Swag annotations in `controllers/hosts.go` and `controllers/jobs.go` named non-existent types (`models.Error`, `ErrorResponse`, `JobsResponse`, `JobSignalRequest`, `JobControlResponseSummary`), had malformed `@Param`/`@Failure` lines and wrong `/api/v1` paths, and no general API info existed. Pre-existing. ## Fix -TODO: required for fixed/partial +Errors use `JsonErrStruct` (what `ReturnError` writes), success `JsonStruct`; bodies use `models.JobSubmitRequest`, `JobControlJobIDRequest`, `JobControlRequest`; paths match the routes; `@BasePath /xce/v1` and general info in `cmd/main.go`; `docs/docs.go`, `swagger.json`, `swagger.yaml` regenerated with `swag init -g cmd/main.go -o docs`. Only hosts and jobs handlers are annotated so far. ## Verification -- Test: TODO: name + file; RED before the fix, GREEN after -- Suites: TODO: suite results and build profile -- Live: TODO: soak/deploy run, or n/a and why +- Test: RED: `swag init -g cmd/main.go` failed with `ParseComment error in file .../controllers/hosts.go ... models.Error: cannot find type definition`; GREEN: swag runs cleanly and the spec lists `/hosts`, `/jobs`, `/jobs/create`, `/jobs/{jobids}`, `/jobs/{jobids}/control`, `/jobs/control`; `cmd` (imports docs) tests pass. +- Suites: `xce-build.sh all asan xce.build --incremental` on the merged branch, 2026-10-10: C suites Ok 157 / Fail 0; frontend Go tests ok (the build now also runs apiserver cmd, middleware, models and router) +- Live: n/a: unit/fixture-level verification; no live deployment was run ## References -- Related: none +- Fix commits: `f7fef278` (on `scheduler-framework`) +- Found as a follow-up of the 2026-10-10 P0 fixes (bug-1936..1957) diff --git a/.wolf/bugs/bug-1967.md b/.wolf/bugs/bug-1967.md new file mode 100644 index 00000000..3392ba79 --- /dev/null +++ b/.wolf/bugs/bug-1967.md @@ -0,0 +1,70 @@ +--- +id: "bug-1967" +status: "open" +status_updated: "2026-10-10" +status_note: "newly logged" +component: "scheduler" +severity: "wrong-result" +timestamp: "2026-10-10T10:47:20Z" +last_seen: "2026-10-10T10:47:20Z" +occurrences: 1 +files: ["src/scheduler/src/plugins/limit.c"] +tags: ["scheduler", "limit", "config"] +related: ["bug-1964", "bug-1685"] +--- + +# bug-1967: xce_scheduler ignores limit changes in a re-pushed config: seeding stops at the first existing name + +## Description +**Observed:** (source reading, 2026-10-10; reported by the bug-1964 fix) +`limit_seed_config_impl()` (`src/scheduler/src/plugins/limit.c`) builds each +configured limit and adds it with `key_add(privdata->limits, limit->name, ...)`, +returning at the first failure: +```c +rc = key_add(privdata->limits, (void*)limit->name, (void*)limit); +... +if (rc < 0) { + ... /* bug-1685: the kvmgr already freed `limit` */ + return rc; +} +``` +On a later config push (live config upgrade, or a mid-replay CONFIG_PUSH in +the simulator since bug-1964) every limit name already exists, so the first +`key_add` fails with `-EEXIST`: no limit is updated, a limit whose value or +scope changed keeps its old definition, and limits listed after an existing +one are never added. + +**Expected:** a config push replaces the configured limits: changed limits +take their new definition, new ones are added, removed ones disappear +(their usage instances handled consistently). +**Impact:** wrong result: limit changes made by a live config upgrade do not +take effect until the scheduler restarts; live and simulator agree (both +wrong). +**Found by:** fix of bug-1964 (sim limit config push), 2026-10-10. +## Environment +- OS: no /etc/os-release; Docker container; NixOS kernel; kernel Linux 6.18.33 x86_64 +- libc: glibc 2.42-47 (interpreter /nix/store/wb6rhpznjfczwlwx23zmdrrw74bayxw4-glibc-2.42-47/lib/ld-linux-x86-64.so.2) +- Toolchain: gcc (GCC) 15.2.0, meson 1.9.1 +- Build: asan (xce.build) +- Code: scheduler-framework @ f3898457-dirty +- Deployment: any scheduler (live config upgrade) and `xce_scheduler --sim` replays with a config push + +## Reproduction +1. Configure limit `L` (value 4) and start the scheduler. +2. Push a config where `L` has value 2 and a new limit `M` follows it. +3. The scheduler still enforces `L` = 4 and has no `M` (REPL `info limit all`). +- Frequency: deterministic (code path) +- Evidence: the code quoted in Description; no logs kept. +## Root cause +TODO: required for fixed/partial + +## Fix +TODO: required for fixed/partial + +## Verification +- Test: TODO: name + file; RED before the fix, GREEN after +- Suites: TODO: suite results and build profile +- Live: TODO: soak/deploy run, or n/a and why + +## References +- Related: none