Closed rksatyam closed 7 years ago
@benheilman any thoughts? This move of the signal handling makes sense
@shortdudey123: Yeah reviewed this with @rksatyam the original patch seems to have been incorrect and this seems like the right way to achieve what the original patch was meant to do.
Sounds good, thats what it looked like
@shortdudey123, Thank you very much for the quick accept.
You are welcome :)
https://github.com/python-diamond/Diamond/commit/7db25796a9c20d81749a76eacf910b371e6652bb caused the regression where the main diamond process stopped handling SIGINT as intended.
The diff want to stop signal handling for the collector processes but was doing it in the collector constructor. The constructor is not the right place to reset as it is really constructed in the main process. Because of this, the signal handler of the main process was getting reset and the signal handler of the main process wasn't really getting called.
Deployed and tested that SIGINT on the main diamond process really kills all the child processes.