I've found about every 24 hours or so pushpoold will "lock up" and stop responding to everything. a killall -HUP doesn't help, I have to kill it and restart it and then everything works fine again. It's getting really bad. Does anyone have any tips?
Check "netstat -au" and see how many TIME_WAIT handles you have open. If that number exceeds your ufile limit, that would be why.
However, in my case, I implemented a patch to fix that problem and I'm still seeing it.
I think the RPC multithreading has some issues - I've observed that every once in awhile, it matches the wrong response to an outgoing request.