Race condition in SshRequestHandler when connection is closed on the server side about trigger

[mwarning]

Thank you for the analysis. Should the exception simple been removed then? I can make a new release once an issue is fixed.

[s3lph]

Removing the exception indeed solves this issue. However, while digging into the code a bit further, I noticed an additional, related issue: session.execCommand is non-blocking. So when read is called after execCommand, there is no guarantee that the entire stdout has been received (or even produced) yet. This can be reproduced by using this simple script on the server side:

#!/usr/bin/env python3

import sys
import time

print('Hello')
sys.stdout.flush()
time.sleep(3)
print('World')

With this server-side script, Trigger sometimes reads the entire string, but sometimes only reads Hello and then terminates.

To capture the entire stdout, waitForCondition and read must be looped until EOF or CLOSED are encountered.