[01:28:32] <groceryheist>	 0/16
[13:47:51] <elukey>	 joal: o/ - I have attached gdb with python debug symbols (so py-bt was available) to one of the rpki spark python workers, and they seem only super slow in calculating the new list of prefixes matching the ip given in input
[13:48:27] <joal>	 WAT?
[13:49:16] <elukey>	 ah didn't expect you online :D
[13:49:45] <elukey>	 yeah it seems simply super slow, python processes in the workers are consuming a ton of cpu time, and all is spent in the list comprehension
[13:50:50] <elukey>	 yep after a while they are making progress
[13:50:59] <elukey>	 (I am crunching only one hour now)
[13:51:10] <joal>	 this is not cool :S Possibly we're gonna try Faidon improvement hu?
[13:51:56] <elukey>	 I am trying it now :)
[13:52:08] <elukey>	 the ipaddress lib seems the cause of the slowdown
[13:52:22] <elukey>	 Faidon's suggestion though is to use pytricia
[13:52:33] <elukey>	 https://pypi.org/project/pytricia/
[13:52:50] <elukey>	 I didn't even know what a patricia tree was :D
[13:53:03] <elukey>	 he should be packaging the lib in a deb package soon
[13:53:13] <elukey>	 so we'll be able to deploy it to the cluster and use it
[13:55:51] <elukey>	 I'll redo tests on Monday with our version, with the ipaddress version and possibily with pytricia
[14:00:34] <elukey>	 (loggin off, ttl! thanks for the chat :)
[14:07:36] <joal>	 Bye ;)