mirror of
https://github.com/windmill-labs/windmill.git
synced 2026-09-12 00:06:14 +00:00
agent-worker-prefetch-throughput
3
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
  pyranotaandClaude Opus 4.7
|
c5f4028a2c |
docs(benchmarks): fix README — port-forward is wm_sim's job
Earlier README told users to run a second `kubectl port-forward` and hard-coded port 33405. Both wrong: wm_sim up calls portForwardApi() in helm_deploy.ts:213, picks a free local port, prints `[helm] API reachable at http://127.0.0.1:<port>`, and keeps the forward alive as a child of the wm_sim process. Updated workflow: read the URL out of wm_sim's output, pass it to main.ts as --host. Kept a note about restarting the forward manually when the kubelet drops it under heavy bench load (real failure mode we hit). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
pyranotaandClaude Opus 4.7
|
4356e60eff |
docs(benchmarks): rewrite sim/README + link from benchmarks/README
sim/README.md was last touched May 27 and described the pre-minikube docker/podman-based "three node modes" design with k8s mode listed as "planned for M5". The reality is now k8s-only via minikube + helm, and the workflow / config-files / outputs all changed. New sim/README.md covers: - The actual `wm_sim up` CLI form against the external windmill-helm-charts chart and smoke.yaml / local.yaml overlays. - Two-shell workflow: wm_sim up in one, port-forward + main.ts in the other, report at reports/<timestamp>/dashboard.svg. - Measurement subsystem table: per-pod CPU/mem sampler DS, JSONL pollers (pod_timeline, oom, pg_latency, pg_conn, node_load), throughput capture, failed-jobs fetch, pgBadger. - The three reliability fixes we leaned on this session and why they matter: sampler host-log file (kubelet log rotation), readiness rollout-complete check (cgroup_mutex starvation), procs_running for oversaturation (not load1 / D-state overcount). - Dashboard panel index + the shared bench_start_ms x-axis origin. - Workload catalog (io_4phase + flood variants + ops_day + cpu). - Agent worker setup with the Secret-based JWT. - Test entry points (util_metrics_test, util_panel_snapshot_test). - Operational notes (port-forward fragility, minikube stop safety). - File-by-file code layout under sim/. benchmarks/README.md gains a "Cluster benchmarks — sim mode" section linking to sim/README.md plus a quick-start snippet so the path from the legacy benchmark suite docs into the cluster workflow is obvious. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
pyranotaandClaude Opus 4.7
|
74b662d8de |
feat(benchmarks): k8s sim mode + util-group dashboard + reliability fixes
Stand up a minikube-backed simulation subsystem for benching Windmill under realistic multi-node load, with a per-bench measurement pipeline and a dashboard renderer that consolidates throughput, queue depth, per-node CPU, PG latency/conns, OOM events, and per-node CPU-util-vs-oversaturation into one SVG report. Sim infrastructure (sim/): - k8s_provisioner: minikube up + heterogeneous node sizing from topology JSON - helm_deploy: helm install Windmill with smoke.yaml + local.yaml overlays - image_cache: pre-load required images so bench bringup is offline-safe - toxiproxy_k8s: per-node toxiproxy DaemonSet for cross-node latency injection - cpu_sampler_k8s: privileged DS reading per-cgroup cpu.stat at 10Hz, dual- writes to stdout AND a host-mounted log file (/var/log/wm-sim-cpu-sampler/ sampler.tsv) so heavy benches no longer lose early samples to kubelet log rotation - pg_logging: ALTER SYSTEM + SIGHUP to enable verbose PG logging without restart - pgbadger: post-bench PG log analysis HTML report - readiness: pre-bench cluster health check (samplers stable ≥30s, workers ready, PG responsive, queue empty, **deploy.status rollout-complete**) — the rollout-complete check catches mid-rolling-update fires that previously starved m04's sampler under cgroup_mutex contention Per-bench JSONL pollers, started/finalized alongside the bench loop: - pod_timeline: 1Hz workers-per-node Ready counts (used for the workers panel) - oom_poller: live OOM event capture (kernel + kubelet evictions + cgroup) - pg_latency_poller: 4Hz psql \\timing on SELECT 1 vs kubectl-exec roundtrip - pg_conn_poller: 1Hz pg_stat_activity by state (active/idle/idle_in_xact) - node_load_poller: 2Hz /proc/loadavg + /proc/stat procs_running per node Dashboard renderer (sim/render_report.ts + graph.ts): - Util group: one panel per node with translucent orange oversaturation area BEHIND solid blue CPU-util area, 100% reference line, phase-boundary verticals. cols:2 grid wraps after 2 panels per row. - PG node tinted with [PG] flag in legend across the dashboard. - Phase-boundary verticals + push-window shaded zones layered consistently. - All x-axes switched from wall-clock HH:MM to relative seconds-from-bench- start. Shared origin sourced from meta.json's bench_start_ms so 0s on every panel = the same wall-clock moment (previously each chart picked its own earliest sample as origin, causing drift between panels). Oversaturation metric, with explicit fallback: - Primary: (procs_running - ncpu) / ncpu × 100 — true CPU run-queue pressure. - Fallback to load1 when procs_running is missing (older reports). - load1 overcounted previously because it includes uninterruptible D-state procs (PG backends in disk I/O, cgroup_mutex waits), inflating "saturation" by 5-10x under load. - Pure helper extracted to sim/util_metrics.ts; 8 unit tests cover the procs_running > load1 preference, the clamp-at-zero, invalid-ncpu cases. Sampler reliability: - HostPath log file in addition to stdout so the bench's scp-based collector bypasses kubelet log rotation entirely. - main.ts truncates the host log file on every node before pushers start (parallel ssh, best-effort) so it doesn't grow unbounded across runs. - Collector falls back to kubectl-logs when scp fails for any node. Workloads (workloads/): - io_4phase: four-phase IO step (idle → 2.5s → 500ms → 150ms jobs) - io_150ms_flood / io_300ms_flood / io_1s_flood / io_2s_flood: single-phase flood configs to isolate the worker-host CFS context-switch storm vs PG contention regime - burst, ops_day, cpu_*, etc. for other scenarios Tests: - sim/util_metrics_test.ts — 8 cases for computeOversatPct - sim/util_panel_snapshot_test.ts — 5 assertions guarding util-panel SVG invariants (orange behind blue, 100% ref line, relative-time ticks NOT wall-clock, phase-boundary verticals, shared-origin override) Helm values: - sim/values/smoke.yaml — bench-tuned: workers w/ no CPU limit & low mem request, PG w/ 3-core request + wm-critical priorityClass + oomImmune + maxConnections, app w/ wm-critical + oomImmune + no resource limits. - sim/values/local.example.yaml — template for the gitignored local.yaml that carries the EE license key. - Depends on the wm-critical PriorityClass + oomImmune + maxConnections knobs landing in windmill-helm-charts (separate PR). graph.ts additions: - areaFills param: ordered list of per-kind translucent area fills drawn before lines, used by the util panel for orange-behind-blue layering - lineColorOverrides: pin per-kind line colors so oversaturation reliably renders orange regardless of d3 ordinal-color insertion order - highlightKindToken: substring-match flag for the PG-node tint in Node CPU - xRelativeOriginMs: shared bench-start origin for the relative-time x-axis - DataPointMulti is now exported for downstream tests Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |