HRPTs, MUD Stability and Performance

[Xygax]

As most of you know, we held an HRPT last night.  I thought you might be interested to know some statistics about the event:

At approximately 8:40 EDT, we had 108 players and 11 staff members logged in.  It doesn't look like we ever went too far over 20-30KB/sec upstream throughput, and ginka's CPUs don't appear to have even broken a sweat.  We fluctuate between 0 and 50% usage of one CPU most of the time; that doesn't seem to have been impacted much by player population.

The game is currently sitting at around 500MB total memory usage (pushed up by the concurrent population from last night; Linux does not reclaim memory allocated to a process -- it can go back into the malloc arena to be re-allocated again, but it is not yielded back to the OS -- unless some trickery is performed, which we don't do.  We also have a not-so-small known memory leak, which should be fixed with the next boot.)  Ginka has 1GB of RAM and 2GB of swap memory available, total, so we're using about half of the "real" memory right now for just the game.

The mud was last booted on Wednesday, Jun 21 at about 9:20AM EDT and has not crashed or been rebooted since then.

Thanks,
-- X