fix(ci): raise AVM check-circuit per-tx timeout to 180s

[AztecBot]

What

The nightly/next AVM Circuit Inputs Collection and Check workflow's avm-check-circuit job has been failing since #23644 (chore: replace bb_cli avm_prove and avm_check_circuit). It first failed on run #1720 (commit 49f6ee5) and again on the scheduled run #1721 (389dc29).

Root cause

Of the ~785 dumped AVM circuit inputs checked per run, exactly one times out: the e2e_multiple_blobs tx 0x03713ef6…. Its check log shows:

• Simulate + APP_LOGIC: ~2s
• Generating trace: ~23s (the bulk)
• Running check (with skippable) circuit over 700560 rows — killed at the 30s mark by timeout -v 30s, exit code 124

Every other tx finishes in 5–6s. The single timeout makes GNU parallel halt (--halt now,fail=1) and the job exits 124.

Each check runs isolated with CPUS=2 (docker --cpus=2), so the 700k-row e2e_multiple_blobs circuit needs ~45s of throttled CPU for trace generation plus the check — just over the existing TIMEOUT=30s. The refactor in #23644 pushed this largest tx across that tight boundary. The per-check budget, not the proving logic, is the problem; the code comment already anticipated this case ("transactions could need more CPU and MEM than we allocate by default … they might start timing out").

Fix

Raise the per-check TIMEOUT from 30s to 180s in avm_check_circuit_cmds (yarn-project/end-to-end/bootstrap.sh). This gives ~4x headroom over the observed ~45s need for the largest tx while staying under the 300s slow-job warning threshold in parallelize. Resource allocation (CPUS=2, MEM=8g) is unchanged to preserve the balanced 64-jobs × 2-CPU = 128-core model and avoid oversubscription; memory peaked at ~3.9 GiB against the 8 GiB limit, so it has ample headroom. The small/fast txs are unaffected.

CI run that surfaced this: https://github.com/AztecProtocol/aztec-packages/actions/runs/26674308080

---

Created by claudebox · group: slackbot