frozenleaves/pytorch
1
0
Fork 0
mirror of https://github.com/pytorch/pytorch.git synced 2025-11-04 08:00:58 +08:00
Code Issues Packages Projects Releases Wiki Activity

[BE] minor logging cleanup in distributed (#122921)

Browse Source 
Summary:
    Minor logging cleanup in distributed library
    1. Don't use "f" formatted strings - address linter issues.
    2. Nits: Make use of unused `e` (error) in a few logs.
    3. Change info->debug as asked in issue #113545
    4. Nit: rename log -> logger in a few files for consistency
    5. Fix a linter error.

    Test Plan:
    1. Local build passes.
    2. Linter is happy.

    Reviewers: wanchaol

Pull Request resolved: https://github.com/pytorch/pytorch/pull/122921
Approved by: https://github.com/wanchaol
This commit is contained in:
Chirag Pandya
2024-03-29 03:33:58 +00:00
committed by PyTorch MergeBot
parent 6a45809580
commit b6201a60c5
25 changed files with 195 additions and 201 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
torch/distributed/elastic/multiprocessing/api.py
View File

@ -38,7 +38,7 @@ IS_WINDOWS = sys.platform == "win32"
IS_MACOS = sys.platform == "darwin"
log = logging.getLogger(__name__)
logger = logging.getLogger(__name__)
__all__ = [
"DefaultLogsSpecs",
@ -260,7 +260,7 @@ class DefaultLogsSpecs(LogsSpecs):
base_log_dir = log_dir or tempfile.mkdtemp(prefix="torchelastic_")
os.makedirs(base_log_dir, exist_ok=True)
dir = tempfile.mkdtemp(prefix=f"{rdzv_run_id}_", dir=base_log_dir)
log.info("log directory set to: %s", dir)
logger.info("log directory set to: %s", dir)
return dir
def reify(self, envs: Dict[int, Dict[str, str]],) -> LogsDest:
@ -276,7 +276,7 @@ class DefaultLogsSpecs(LogsSpecs):
if nprocs > 0:
global_env = envs[0]
else:
log.warning("Empty envs map provided when defining logging destinations.")
logger.warning("Empty envs map provided when defining logging destinations.")
# Keys are always defined, but values can be missing in unit tests
run_id = global_env.get("TORCHELASTIC_RUN_ID", "test_run_id")
restart_count = global_env.get("TORCHELASTIC_RESTART_COUNT", "0")
@ -355,7 +355,7 @@ class DefaultLogsSpecs(LogsSpecs):
error_file = os.path.join(clogdir, "error.json")
error_files[local_rank] = error_file
log.info("Setting worker%s reply file to: %s", local_rank, error_file)
logger.info("Setting worker%s reply file to: %s", local_rank, error_file)
envs[local_rank]["TORCHELASTIC_ERROR_FILE"] = error_file
return LogsDest(stdouts, stderrs, tee_stdouts, tee_stderrs, error_files)
@ -692,7 +692,7 @@ class MultiprocessContext(PContext):
failed_proc = self._pc.processes[failed_local_rank]
error_filepath = self.error_files[failed_local_rank]
log.exception(
logger.exception(
"failed (exitcode: %s)"
" local_rank: %s (pid: %s)"
" of fn: %s (start_method: %s)",
@ -724,7 +724,7 @@ class MultiprocessContext(PContext):
return
for proc in self._pc.processes:
if proc.is_alive():
log.warning("Closing process %s via signal %s", proc.pid, death_sig.name)
logger.warning("Closing process %s via signal %s", proc.pid, death_sig.name)
try:
os.kill(proc.pid, death_sig)
except ProcessLookupError:
@ -739,7 +739,7 @@ class MultiprocessContext(PContext):
proc.join(time_to_wait)
for proc in self._pc.processes:
if proc.is_alive():
log.warning(
logger.warning(
"Unable to shutdown process %s via %s, forcefully exiting via %s",
proc.pid, death_sig, _get_kill_signal()
)
@ -823,7 +823,7 @@ class SubprocessContext(PContext):
)
if result.is_failed():
first_failure = min(result.failures.values(), key=lambda f: f.timestamp)
log.error(
logger.error(
"failed (exitcode: %s)"
" local_rank: %s (pid: %s)"
" of binary: %s",
@ -848,7 +848,7 @@ class SubprocessContext(PContext):
return
for handler in self.subprocess_handlers.values():
if handler.proc.poll() is None:
log.warning(
logger.warning(
"Sending process %s closing signal %s", handler.proc.pid, death_sig.name
)
handler.close(death_sig=death_sig)
@ -865,7 +865,7 @@ class SubprocessContext(PContext):
pass
for handler in self.subprocess_handlers.values():
if handler.proc.poll() is None:
log.warning(
logger.warning(
"Unable to shutdown process %s via %s, forcefully exiting via %s",
handler.proc.pid, death_sig, _get_kill_signal()
)