mirror of https://github.com/windmill-labs/windmill.git synced 2026-08-19 00:02:03 +00:00

Files

T

History

pyranota fadbd73dc6 perf(agent-workers): two-stage prefetch pipeline for agent throughput

Experimental and env-gated (defaults off — stock behavior unchanged).

Read path is made fully non-blocking on both sides so neither the worker
nor the server waits on the stage behind it:
  Postgres -> [server-side buffer] -> HTTP -> [agent-side buffer] -> worker

- batch pull/complete: N jobs per pull, N completions per flush, so the
  per-job HTTP round-trips and per-job dequeue queries are amortized
- server-side prefetch: a single background puller per job-tag-set runs
  the SKIP LOCKED dequeue ahead of demand and coalesces concurrent agents,
  moving the PG query off the request critical path
- agent-side prefetch: a background refiller keeps the local job buffer
  deep (high low-watermark + several refills in flight) so the worker
  loop pops without a network call
- empty-pull backoff fix: don't apply the idle long-sleep to a transient
  empty pull from a prefetching consumer
- AGENT_PROF: per-phase profiler + auto-rendered per-job time-budget SVG

Single agent goes from ~2.5k to ~125k+ j/s. Multi-agent collapse is
reproduced but its root cause is still open (not established to be PG).
See benchmarks/agent_worker_scaling.md.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

2026-06-20 22:09:53 +02:00

nsjail

feat: sandboxed daemonless container runtime via '# sandbox <image>' (#9453 )

2026-06-05 08:35:51 +00:00

src

perf(agent-workers): two-stage prefetch pipeline for agent throughput

2026-06-20 22:09:53 +02:00

build.rs

refactor: isolate deno_core into windmill-runtime-nativets subcrate (#7848 )

2026-02-08 09:16:37 +00:00

Cargo.toml

perf(agent-workers): two-stage prefetch pipeline for agent throughput