Skip to content

continuous

continuous #774

Workflow file for this run

# Continuous Testing
#
# Run 50 times per day, split over two runs to keep individual run times
# lower so as not to hit action timeout conditions.
name: continuous
on:
schedule:
- cron: "0 10 * * *" # Every day at 10:00 UTC
- cron: "30 23 * * *" # Every day at 23:30 UTC
workflow_dispatch: # Allows you to run this workflow manually from the Actions tab
jobs:
preqs:
runs-on: macos-latest
steps:
- uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # ratchet:actions/checkout@v6
# Note: Using Bazel flags documented here to reduce memory footprint:
# https://bazel.build/advanced/performance/memory
#
# The --noexperimental_collect_* flags disable bazel-internal metrics
# collectors that have no value here (we don't capture profiles or
# consume BEP from runner output). One of them (system network usage)
# was observed crashing the bazel server with `sysctl: Cannot allocate
# memory` on the 7GB macos runners. If a future Bazel bump fails with
# "Unrecognized option" on any of these, the `experimental_` prefix
# has likely been dropped or the flag renamed; run `bazel help build`
# against the new version to reconcile.
- name: Check for flaky tests
run: |
bazel test \
--test_strategy=exclusive \
--test_output=errors \
--runs_per_test 25 \
--discard_analysis_cache \
--notrack_incremental_state \
--nokeep_state_after_build \
--noexperimental_collect_system_network_usage \
--noexperimental_collect_load_average_in_profiler \
--noexperimental_collect_resource_estimation \
--noexperimental_collect_local_sandbox_action_metrics \
--show_progress_rate_limit=5 \
--curses=no \
--color=no \
--define=SANTA_BUILD_TYPE=adhoc \
-t- \
:unit_tests