Skip to content

change: [smdataparallel] better messages to establish the SSH connection between workers #103

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Apr 12, 2021
Merged

change: [smdataparallel] better messages to establish the SSH connection between workers #103

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
0dcd738
change: [smdataparallel] better messages for to establish the SSH con…
karan6181 Apr 8, 2021
1605c6a
python timeout.timeout raises TimeoutError
karan6181 Apr 8, 2021
886a345
Added detailed error message
karan6181 Apr 8, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
19 changes: 12 additions & 7 deletions src/sagemaker_training/smdataparallel.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -86,11 +86,17 @@ def _wait_for_workers(self): # type: () -> None
logger.info("Waiting for MPI workers to establish their SSH connections")

workers = [host for host in self._hosts if host != self._master_hostname]
with timeout.timeout(seconds=self.timeout_in_seconds):
for host in workers:
while not _can_connect(host):
time.sleep(self._interval)
logger.info("Worker %s available for communication", host)
try:
with timeout.timeout(seconds=self.timeout_in_seconds):
for host in workers:
while not _can_connect(host):
time.sleep(self._interval)
logger.info("Worker %s available for communication", host)
except timeout.TimeoutError:
logger.exception(
"Connection between the hosts couldn't established. Aborting the training."
)
raise

def _get_mpirun_command(
self,
Expand Down Expand Up @@ -321,8 +327,7 @@ def _can_connect(host, port=22):
logger.info("Can connect to host %s at port %s", host, port)
return True
except Exception: # pylint: disable=broad-except
logger.info("Cannot connect to host %s at port %s", host, port)
logger.exception("Connection failed")
logger.info("Cannot connect to host %s at port %s. Retrying...", host, port)
return False
finally:
client.close()
Expand Down