Open cstanfill opened 13 years ago
If this is a threading issue, perhaps setting the new connections to a clock heartbeat of some kind would help?
Hm. Will look into it. Bug now also manifests itself in the form of the program suddenly exiting without a segfault.
The non-segfault version has been resolved; it was exiting on SIGPIPE; a flag passed to the send command fixes that. Need to catch the error it does return, though. Not sure how to reproduce the segfault now.
The server program occasionally segfaults for no apparent reason. Happens when it is being hammered - running the stresser script with 127.0.0.1 6667 100000 will generally cause it within 10-20 connections. Haven't investigated much.