worker: Do not check if the process is still alive prior to using it.

This might sound strange at first, but the reasoning is this: A worker should never simply exit. Bazel controls the lifetime of its subprocesses, so a worker quitting is considered a failure (also because it cannot be distinguished from a crash).

With that set, we can improve two things:
- Bazel will now only notice that a worker crashed / quit when it tries to use one during a build. This is also the only time when we can print error messages to the user. Earlier we might have noticed that a worker crashed during validation, but had no mechanism to alert the user to this, because this wasn't necessarily during a build.
- This also fixes a race condition where a worker is still alive during validation, then quits, then the WorkerSpawnStrategy tries to send a WorkRequest, which fails, triggering an IOException.

This fixes the flaky test test_worker_restarts_after_exit (which is now called test_build_fails_when_worker_exits).

Part of #2855.

RELNOTES: Bazel will no longer gracefully restart workers that crashed / quit, instead this triggers a build failure.
PiperOrigin-RevId: 154470257

1 files changed, 5 insertions, 24 deletions

diff --git a/src/test/shell/integration/bazel_worker_test.sh b/src/test/shell/integration/bazel_worker_test.sh
index 1f20479e9f..d8877ba8ce 100755
--- a/src/test/shell/integration/bazel_worker_test.sh
+++ b/src/test/shell/integration/bazel_worker_test.sh
@@ -228,41 +228,24 @@ EOF
   assert_equals "1" $work_count
 }
 
-function test_worker_restarts_after_exit() {
+function test_build_fails_when_worker_exits() {
   prepare_example_worker
   cat >>BUILD <<'EOF'
 [work(
   name = "hello_world_%s" % idx,
   worker = ":worker",
-  worker_args = ["--exit_after=2"],
+  worker_args = ["--exit_after=1"],
   args = ["--write_uuid", "--write_counter"],
 ) for idx in range(10)]
 EOF
 
   bazel build :hello_world_1 &> $TEST_log \
     || fail "build failed"
-  worker_uuid_1=$(cat $BINS/hello_world_1.out | grep UUID | cut -d' ' -f2)
-  work_count=$(cat $BINS/hello_world_1.out | grep COUNTER | cut -d' ' -f2)
-  assert_equals "1" $work_count
 
   bazel build :hello_world_2 &> $TEST_log \
-    || fail "build failed"
-  worker_uuid_2=$(cat $BINS/hello_world_2.out | grep UUID | cut -d' ' -f2)
-  work_count=$(cat $BINS/hello_world_2.out | grep COUNTER | cut -d' ' -f2)
-  assert_equals "2" $work_count
-
-  # Check that the same worker was used twice.
-  assert_equals "$worker_uuid_1" "$worker_uuid_2"
+    && fail "expected build to failed" || true
 
-  bazel build :hello_world_3 &> $TEST_log \
-    || fail "build failed"
-  worker_uuid_3=$(cat $BINS/hello_world_3.out | grep UUID | cut -d' ' -f2)
-  work_count=$(cat $BINS/hello_world_3.out | grep COUNTER | cut -d' ' -f2)
-  assert_equals "1" $work_count
-  expect_log "worker .* can no longer be used, because its process terminated itself or got killed"
-
-  # Check that we used a new worker.
-  assert_not_equals "$worker_uuid_2" "$worker_uuid_3"
+  expect_log "Worker process quit or closed its stdin stream when we tried to send a WorkRequest"
 }
 
 function test_worker_restarts_when_worker_binary_changes() {
@@ -533,14 +516,12 @@ EOF
 
   bazel build :hello_world_1 &> $TEST_log \
     || fail "build failed"
-  worker_uuid_1=$(cat $BINS/hello_world_1.out | grep UUID | cut -d' ' -f2)
 
   bazel build :hello_world_2 &> $TEST_log \
     && fail "expected build to fail" || /bin/true
-  worker_uuid_2=$(cat $BINS/hello_world_2.out | grep UUID | cut -d' ' -f2)
 
   expect_log "^---8<---8<--- Start of log, file at /"
-  expect_log "thus dumping its log file for debugging purposes:"
+  expect_log "Worker process did not return a WorkResponse:"
   expect_log "I'm a very poisoned worker and will just crash."
   expect_log "^---8<---8<--- End of log ---8<---8<---"
 }