get_pg_node_state can fail (PID file is empty) #329
Description
https://github.com/dmitry-lipetsk/testgres/actions/runs/21317871218/job/61363311353
________________ TestSet001__start.test_002__wait_false[local2] ________________
[gw1] linux -- Python 3.7.17 /tmp/testgres_venv/bin/python3
self = <tests.units.node.PostgresNode.test_setM001__start.TestSet001__start object at 0x7fb5b7bb5ad0>
node_svc = <tests.helpers.global_data.PostgresNodeService object at 0x7fb5b7a3f990>
def test_002__wait_false(self, node_svc: PostgresNodeService):
assert isinstance(node_svc, PostgresNodeService)
C_MAX_ATTEMPTS = 3
attempt = 0
while True:
assert type(attempt) == int # noqa: E721
assert attempt >= 0
assert attempt <= C_MAX_ATTEMPTS
if attempt == C_MAX_ATTEMPTS:
raise RuntimeError("Node is not started")
attempt += 1
logging.info("------------- attempt #{}".format(attempt))
if attempt > 1:
HelperUtils.PrintAndSleep(5)
with PostgresNodeTestUtils.get_node(node_svc) as node:
assert type(node) == PostgresNode # noqa: E721
node.init()
assert not node.is_started
assert node.status() == NodeStatus.Stopped
node_log_reader = PostgresNodeLogReader(node, from_beginnig=False)
node.start(wait=False)
assert node.is_started
assert node.status() in [NodeStatus.Stopped, NodeStatus.Running]
# Internals
assert type(node._manually_started_pm_pid) == int # noqa: E721
assert node._manually_started_pm_pid == node._C_PM_PID__IS_NOT_DETECTED
logging.info("Wait for running state ...")
try:
PostgresNodeTestUtils.wait_for_running_state(
node=node,
node_log_reader=node_log_reader,
> timeout=60,
)
C_MAX_ATTEMPTS = 3
attempt = 1
node = PostgresNode(name='testgres-9087a3b7-ddd4-433e-bd21-2b58b9bcb415', port=None, base_dir='/tmp/tgsn__ltrmhq7')
node_log_reader = <src.node.PostgresNodeLogReader object at 0x7fb5b6be43d0>
node_svc = <tests.helpers.global_data.PostgresNodeService object at 0x7fb5b7a3f990>
self = <tests.units.node.PostgresNode.test_setM001__start.TestSet001__start object at 0x7fb5b7bb5ad0>
tests/units/node/PostgresNode/test_setM001__start.py:130:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
tests/helpers/pg_node_utils.py:168: in wait_for_running_state
s = node.status()
_ = None
__class__ = <class 'tests.helpers.pg_node_utils.PostgresNodeUtils'>
node = PostgresNode(name='testgres-9087a3b7-ddd4-433e-bd21-2b58b9bcb415', port=None, base_dir='/tmp/tgsn__ltrmhq7')
node_log_reader = <src.node.PostgresNodeLogReader object at 0x7fb5b6be43d0>
timeout = 60
src/node.py:930: in status
x = self._get_node_state()
self = PostgresNode(name='testgres-9087a3b7-ddd4-433e-bd21-2b58b9bcb415', port=None, base_dir='/tmp/tgsn__ltrmhq7')
src/node.py:939: in _get_node_state
self.utils_log_file
self = PostgresNode(name='testgres-9087a3b7-ddd4-433e-bd21-2b58b9bcb415', port=None, base_dir='/tmp/tgsn__ltrmhq7')
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
os_ops = <testgres.operations.local_ops.LocalOperations object at 0x7fb5b85d4e50>
bin_dir = '/usr/local/bin', data_dir = '/tmp/tgsn__ltrmhq7/data'
utils_log_file = '/tmp/tgsn__ltrmhq7/logs/utils.log'
def get_pg_node_state(
os_ops: OsOperations,
bin_dir: str,
data_dir: str,
utils_log_file: typing.Optional[str],
) -> PostgresNodeState:
assert isinstance(os_ops, OsOperations)
assert type(bin_dir) == str # noqa: E721
assert type(data_dir) == str # noqa: E721
assert utils_log_file is None or type(utils_log_file) == str # noqa: E721
_params = [
os_ops.build_path(bin_dir, "pg_ctl"),
"-D",
data_dir,
"status",
]
status_code, out, error = execute_utility2(
os_ops,
_params,
utils_log_file,
verbose=True,
ignore_errors=True,
)
assert type(status_code) == int # noqa: E721
assert type(out) == str # noqa: E721
assert type(error) == str # noqa: E721
# -----------------
if status_code == PG_CTL__STATUS__NODE_IS_STOPPED:
return PostgresNodeState(NodeStatus.Stopped, None)
# -----------------
if status_code == PG_CTL__STATUS__BAD_DATADIR:
return PostgresNodeState(NodeStatus.Uninitialized, None)
# -----------------
if status_code != PG_CTL__STATUS__OK:
errMsg = "Getting of a node status [data_dir is {0}] failed.".format(
data_dir
)
raise ExecUtilException(
message=errMsg,
command=_params,
exit_code=status_code,
out=out,
> error=error,
)
E testgres.operations.exceptions.ExecUtilException: Getting of a node status [data_dir is /tmp/tgsn__ltrmhq7/data] failed.
E Command: /usr/local/bin/pg_ctl -D /tmp/tgsn__ltrmhq7/data status
E Exit code: 1
E ---- Error:
E pg_ctl: the PID file "/tmp/tgsn__ltrmhq7/data/postmaster.pid" is empty
_params = ['/usr/local/bin/pg_ctl', '-D', '/tmp/tgsn__ltrmhq7/data', 'status']
bin_dir = '/usr/local/bin'
data_dir = '/tmp/tgsn__ltrmhq7/data'
errMsg = 'Getting of a node status [data_dir is /tmp/tgsn__ltrmhq7/data] failed.'
error = 'pg_ctl: the PID file "/tmp/tgsn__ltrmhq7/data/postmaster.pid" is empty\n'
os_ops = <testgres.operations.local_ops.LocalOperations object at 0x7fb5b85d4e50>
out = ''
status_code = 1
utils_log_file = '/tmp/tgsn__ltrmhq7/logs/utils.log'
src/utils.py:392: ExecUtilException