fix(ci): raise AVM check-circuit per-test timeout to 120s

[AztecBot]

Problem

The avm-check-circuit job failed on next (run 27215957791, exit code 124 = timeout).

The GitHub job failure is just the symptom; the actual failure was a single AVM check-circuit test hitting its per-test timeout:

FAILED: bb-avm avm_check_circuit -v --avm-inputs .../e2e_multiple_blobs/avm-circuit-inputs-tx-0x017d645b...bin (35s) (code: 124)
run_test_cmd '...ISOLATE=1:TIMEOUT=30s:NAME=avm_cc_e2e_multiple_blobs_0x017d645b ...'

This is not a constraint failure — the circuit is valid. The individual test log shows it was simply killed mid-check after exceeding the 30s budget:

• Trace generation: ~16s
• Running check (with skippable) circuit over 700560 rows started, then timeout: sending signal TERM at ~30s

Root cause

avm_check_circuit_cmds runs every dumped tx with TIMEOUT=30s and the default 2 CPUs / 8g. Most e2e txs produce small AVM traces that clear check-circuit in 4–8s, but a few (notably e2e_multiple_blobs, ~700k circuit rows) need ~30–40s for trace generation plus check-circuit at 2 CPUs. The 30s budget was right on the edge, so this tx flakes. The triggering commit (#23963, a CI spot-instance change) is unrelated to the AVM — this is a pre-existing borderline-timeout test, not a regression.

Change

Raise the per-test timeout 30s → 120s in avm_check_circuit_cmds.

A longer timeout — rather than raising CPUS — is the targeted fix: the parallel scheduler acquires CPUS cores per test from a global semaphore, so bumping CPUS from 2 to 4 would halve concurrency (96 → 48 on a 192-core runner) for every tx, penalizing the hundreds of small txs to help a handful of large outliers. A longer timeout only lets the rare large tx hold its slot a little longer, with zero impact on the common case. 120s gives ~3x headroom over the observed ~36s worst case.

Validation

The failing run shows the test executing correctly (valid trace, valid in-progress check) and being killed purely by the wall-clock timeout, so additional time resolves it. The change is a single comment + timeout-value edit in the CI test-command generator.

---

Created by claudebox · group: slackbot

[AztecBot]

Flakey Tests

🤖 says: This CI run detected 1 tests that failed, but were tolerated due to a .test_patterns.yml entry.

\033FLAKED\033 (8;;http://ci.aztec-labs.com/1ae79e4b27d3d437�1ae79e4b27d3d4378;;�):  yarn-project/end-to-end/scripts/run_test.sh simple src/e2e_p2p/sentinel_status_slash.parallel.test.ts "slashes the proposer with INACTIVITY when checkpoint validation records invalid" (237s) (code: 0) group:e2e-p2p-epoch-flakes