amadan
gate PROBE (revert me): skip internal/console entirely to see whether the runner's TERM follows the console tests
Ten CI runs on this branch since Task 7 ended with the step killed by SIGTERM about 35 s after the eleven test shards started, with the gate's own TERM trap never reached. This run tests nothing in internal/console (no shards, not in rest); if it survives, the console tests trigger the kill and the next probes bisect them; if it dies the same way, they do not. Reverted by the next commit once the run has reported.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
1 file changed,
+6
−2
.amadan/test-gate.sh+6 −2
CI failed — run details
CI log
=== gate === .amadan/test-gate.sh gate: 2026-10-24T00:15:13Z on aarch64, 32 cpus, 62062 MB available, go1.26.8 linux/arm64 gate: +0s start launchctx gate: +0s start vet gate: +0s start gofmt gate: +1s start rest gate: +13s start root@0 gate: +13s start root@1 gate: +13s start root@2 amadan: step gate failed: signal: terminated amadan: job failed at step gate — its output is above, not at the tail of this log
diff --git a/.amadan/test-gate.sh b/.amadan/test-gate.sh| index c1fe71009..2a5be5107 100755 |
| --- a/.amadan/test-gate.sh |
| +++ b/.amadan/test-gate.sh |
| @@ -33,7 +33,11 @@ |
| # next) can surface here. That is a bug in the test, not in the split. |
| set -eu |
| -: "${TEST_SPLIT=./internal/console=8 .=3}" |
| +: "${TEST_SPLIT=.=3}" |
| +# PROBE (revert me): internal/console is not tested at all in this run, to |
| +# learn whether the runner's TERM (ten runs, ~35 s after the shards start) |
| +# follows the console tests. Nothing else changes. |
| +PROBE_SKIP="github.com/carlosframework/platform/internal/console" |
| weights=$(pwd)/.amadan/test-weights.tsv |
| # A shard count that is not a whole number of at least 1 would take a |
| @@ -179,7 +183,7 @@ if [ -z "$all" ]; then |
| fi |
| rest="" |
| for p in $all; do |
| - case " $split_paths " in |
| + case " $split_paths $PROBE_SKIP " in |
| *" $p "*) ;; |
| *) rest="$rest $p" ;; |
| esac |