community 
How to implement timeout?
How do we implement a timeout?
You know how in the old authoring client, each step could have a timeout?
In Python, how would we do that? Specifically, 2 cases:
1. If we do an “ExecuteCommand” from the SETUP or TEARDOWN script to a driver command – how do we set a timeout where if it doesn’t return after a set amount of time, it terminates it and then raises an exception?
2. In a driver itself, if for example, we try to call a function and we want to set a timeout for the function to return?
Hi @mchua,
First, make sure that your driver commands support cancellation (as much as you can), so that in a case of getting stopped (either from the portal or with the API), the driver will be able to stop at the right time or roll back parts of what it did if it's required. You can learn more about cancellation and termination support in the dev guide.
You should also check in the driver itself, why a command should take so long, sometimes, you can set the timeout when you call the internal modules (like setting a timeout on a telnet/ssh/rest calls).
Next, if you have a function that you can poll and check the progress/result and decide if you'd like to stop after a certain timeout, then you can do a while loop that validates that:
def _run_something_with_timeout(arg1, arg2, timeout=10, cancellation_context)
start_time = time.time()
data = self._get_data(arg1, arg2)
while not data:
if time.time() - start_time >= timeout:
raise TimeoutError('Timeout: Waiting for data')
# check if CloudShell sent a cancel update
if cancellation_context.is_cancelled:
raise CommandCancelledException("Command was Cancelled")
time.sleep(5)
data = self._get_data(arg1, arg2)
return data
def my_driver_command(self, context, cancellation_context):
try:
data = _run_something_with_timeout('a', 'b', timeout=30, cancellation_context)
except TimeoutError as te:
print 'timeout'
return ''
except CommandCancelledException as cce:
return 'command cancelled'
If you'd like to do something similar from the orchestration script, that will execute the command and will poll for its status, you'll need to use the Sandbox API, which allows you to start a command, get an execution id, poll the status, and if needed, call the stop execution (with the cloudshell-automation-api, you can either ExecuteCommand or EnqueueCommand (without knowing its status)). There is a python sandbox api wrapper that can get you started (it's not covering executing/stopping commands yet, but should be hard to add them).
Yaniv
