Overview

7 days30 days

Latest CI Pipeline Executions

Fuzzy

Succeeded
35650
9da97d04 chore(misc): bump nx.json bust to force e2e re-run
by Jason Jean
Failed
35650
Flaky task→
bb060073 chore(misc): bump nx.json bust to force e2e re-run
by Jason Jean
Succeeded
35650
bb060073 chore(misc): bump nx.json bust to force e2e re-run
by Jason Jean
Failed
35650
Flaky task→
b7e42f4e chore(misc): bump nx.json bust to force e2e re-run
by Jason Jean
Failed
35650
Environment issue→
3b95887f chore(core): log plugin iteration in createProjectConfigurationsWithPlugins
by Jason Jean
Canceled
35650
3b95887f chore(core): log plugin iteration in createProjectConfigurationsWithPlugins
by Jason Jean
Failed
35650
Flaky task→
3b95887f chore(core): log plugin iteration in createProjectConfigurationsWithPlugins
by Jason Jean
Failed
35650
44b6eb3f chore(core): per-stage merge dumps under NX_DAEMON_DEBUG_WATCHER Spread-flake unit harness (6 levels) proves the merge stack itself is not the bug — all six harness tests pass even with the intermediate-default overlay and validateAndNormalizeProjectRootMap applied. Harness 6 does surface that an incompatible-target overlay silently drops props like inputs, but that's not the e2e failure mode (build itself is missing). Adds [debug-merge] dumps at five points in the production pipeline: after-specified, after-target-defaults, intermediate, after-intermediate-overlay, after-apply-substitutions, after-normalize. One CI run on the flake will pin which stage drops libs/lib<N>.targets.build. Bumps nx.json bust to force the e2e to re-run.
by Jason Jean
Failed
35650
Flaky task→
44b6eb3f chore(core): per-stage merge dumps under NX_DAEMON_DEBUG_WATCHER Spread-flake unit harness (6 levels) proves the merge stack itself is not the bug — all six harness tests pass even with the intermediate-default overlay and validateAndNormalizeProjectRootMap applied. Harness 6 does surface that an incompatible-target overlay silently drops props like inputs, but that's not the e2e failure mode (build itself is missing). Adds [debug-merge] dumps at five points in the production pipeline: after-specified, after-target-defaults, intermediate, after-intermediate-overlay, after-apply-substitutions, after-normalize. One CI run on the flake will pin which stage drops libs/lib<N>.targets.build. Bumps nx.json bust to force the e2e to re-run.
by Jason Jean
Canceled
35650
b0c62cdd chore(misc): bump nx.json bust to force e2e re-run Bumps bust 3222 -> 3223 so Nx Cloud re-runs the e2e suites instead of replaying cached results. Needed so the watcher diagnostics on this branch get real execution data, including from any flaky tests we want to observe.
by Jason Jean
Failed
35650
Fix applied→
b0c62cdd chore(misc): bump nx.json bust to force e2e re-run Bumps bust 3222 -> 3223 so Nx Cloud re-runs the e2e suites instead of replaying cached results. Needed so the watcher diagnostics on this branch get real execution data, including from any flaky tests we want to observe.
by Jason Jean
Failed
35650
Fix ready→
5bdd81bd chore(core): log event age + force-flush timing in NX_DAEMON_DEBUG_WATCHER Adds three measurements behind the existing NX_DAEMON_DEBUG_WATCHER=1 gate to help localize where the spread-test flake's latency lives: 1. Event age — in ingest_event, log every event with `age_ms = now() - mtime`. Surfaces upstream delivery delay (kernel inotify buffer → notify-crate worker → notify_rx) that's invisible from inside the pipeline. A large age means the file was written well before our code saw the event. 2. Force-flush START/END timestamps + duration_ms — wraps the force_flush_rx arm so we can see how long the handler itself took, separate from the upstream delay. 3. recv_timeout(5ms) outcome — Ok/Timeout/Disconnected so we can tell whether the 5ms in-flight wait caught anything or sat empty. All gated on NX_DAEMON_DEBUG_WATCHER=1; production behaviour unchanged. Wall- clock timestamps are formatted HH:MM:SS.mmm in UTC to line up with the JS daemon log's existing prefix when correlating after a flake.
by Jason Jean
Failed
35650
5bdd81bd chore(core): log event age + force-flush timing in NX_DAEMON_DEBUG_WATCHER Adds three measurements behind the existing NX_DAEMON_DEBUG_WATCHER=1 gate to help localize where the spread-test flake's latency lives: 1. Event age — in ingest_event, log every event with `age_ms = now() - mtime`. Surfaces upstream delivery delay (kernel inotify buffer → notify-crate worker → notify_rx) that's invisible from inside the pipeline. A large age means the file was written well before our code saw the event. 2. Force-flush START/END timestamps + duration_ms — wraps the force_flush_rx arm so we can see how long the handler itself took, separate from the upstream delay. 3. recv_timeout(5ms) outcome — Ok/Timeout/Disconnected so we can tell whether the 5ms in-flight wait caught anything or sat empty. All gated on NX_DAEMON_DEBUG_WATCHER=1; production behaviour unchanged. Wall- clock timestamps are formatted HH:MM:SS.mmm in UTC to line up with the JS daemon log's existing prefix when correlating after a flake.
by Jason Jean
Failed
35650
Fix available→
90b32ff4 chore(core): add NX_DAEMON_DEBUG_WATCHER for diagnosing spread flake Adds opt-in logging at the suspected hot paths for the daemon watcher race, gated by NX_DAEMON_DEBUG_WATCHER=1: - routeWorkspaceChanges logs every event batch it receives plus any paths dropped (dir, stat error). - getCachedSerializedProjectGraphPromise logs its needsRecompute decision with the current collected* contents. - processCollectedUpdatedAndDeletedFiles logs the configHash compare that decides between full re-scan and incremental update. - Native register_and_backfill_new_dirs (Rust) prints which dirs got watches and which files were backfilled. The spread test now sets the flag and copies the daemon log to /tmp/spread-debug-logs/<test-name>.log after each test so CI keeps the diagnostic artifact across runs. No production behaviour change when the flag is unset.
by FrozenPandaz
Failed
35650
Fix applied→
90b32ff4 chore(core): add NX_DAEMON_DEBUG_WATCHER for diagnosing spread flake Adds opt-in logging at the suspected hot paths for the daemon watcher race, gated by NX_DAEMON_DEBUG_WATCHER=1: - routeWorkspaceChanges logs every event batch it receives plus any paths dropped (dir, stat error). - getCachedSerializedProjectGraphPromise logs its needsRecompute decision with the current collected* contents. - processCollectedUpdatedAndDeletedFiles logs the configHash compare that decides between full re-scan and incremental update. - Native register_and_backfill_new_dirs (Rust) prints which dirs got watches and which files were backfilled. The spread test now sets the flag and copies the daemon log to /tmp/spread-debug-logs/<test-name>.log after each test so CI keeps the diagnostic artifact across runs. No production behaviour change when the flag is unset.
by FrozenPandaz