Well if there is one thing that I've learned about bitcoin mining pools is that they're not exactly easy on server resources and it seems it's quite a challenge to keep a pool running without any problems. Also let's not forget the potential for getting DDoS'ed too!

How soon before long polling?

2011-06-03 19:13:42: Listener for "worker41": 03/06/2011 19:13:42, Problems communicating with bitcoin RPC
2011-06-03 19:13:47: Listener for "worker18": 03/06/2011 19:13:47, Problems communicating with bitcoin RPC
2011-06-03 19:14:36: Listener for "worker27": 03/06/2011 19:14:36, Problems communicating with bitcoin RPC
2011-06-03 19:14:51: Listener for "worker22": 03/06/2011 19:14:51, Problems communicating with bitcoin RPC
2011-06-03 19:15:06: Listener for "worker29": 03/06/2011 19:15:06, Problems communicating with bitcoin RPC
2011-06-03 19:15:53: Listener for "worker39": 03/06/2011 19:15:53, Problems communicating with bitcoin RPC
2011-06-03 19:16:17: Listener for "worker16": 03/06/2011 19:16:17, Problems communicating with bitcoin RPC
2011-06-03 19:17:25: Listener for "worker02": 03/06/2011 19:17:25, Problems communicating with bitcoin RPC

I finally have response from Linode guys, they decided to raise limits for pool significantly. Unfortunately it need balancer restart to take effect, so expect short outage.

This should solve stale rate and RPC errors (but today's errors were caused by database error, which is already solved).

EDIT: Great, looks like that annoying packet loss is solved! (0% in 2000 requests)

By the way, there is another guy with access to pool servers. He's currently learning how to fix some specific problems for times when I'm offline. He is very trustworthy for me and well known on this forum, but he doesn't want to reveal his identity; he simply doesn't have enough time for 'normal agenda' like responding emails, replying on forum and so on. But this independent supervision is definitely making operator's response faster in case of any problem.

Just want to say thank you Slush, the connection stability and overall responsiveness has been greatly increased  
There are still some issues with idle miners, but it is a lot better then before.

same here, that round was unusually low for me. I think maybe the pool had a dc problem.

Ooofah, cluster performance over 700 GH/sec, wasn't it under 500 just a week or so ago?

Damn, I always forget to check if score renormalization finished when there're database problems. I fixed rewards, should be fine now.