The results are in. Using a 750mhz athlon w/ 64mb ram and the same test application as the last stress test report, we see good results. 9 concurrent connections were used in this test.
Mud handled 90 movement commands/second (spread across 9 players) for 1min 7secs. With 'say' commands at the same rate the player threads crased after 45 seconds.
At a lower rate of 22.5 commands/sec the test results became somewhat suspicious: exactly 3 minutes with movement commands and exactly 4 minutes with 'say' commands. I think this may either reflect a problem with my testing application or some kind of leak in the mud code.
All the tests maxed the processor on my 750mhz athlon, but ran with very little memory usage.... read more
About two weeks ago nearly all of the code in RbMUD-0.1 was abandoned by the developers and replaced with a much more elegant object oriented system . The new code features an sql backend and vastly increased flexibility. It is also much easier to work with. Unfortunately, due to the complexity of the style of inheritence we chose to implement and a plethora of database glitches, development has been painful and slow.
With ~1600 new lines of code and a beligerently functional database, the new core is finally starting to take shape. Expect RbMUD-0.2 to include mostly just the features in RbMUD-0.1, from the player standpoint.
In conclusion, to prove the increase of quality over the old system, I plan to release a new set of stress tests on the CVS code. Hopefully we will see a considerable increase in stability and reliability, especially under heavier loads than previously possible.
Big update this time. The 0.1rc3 code is radically improved. The largest addition is prog support. The core of the system has also been shuffled around (optimized & improved), and major bugs in the command handling code have been fixed. The code is now reaching a fairly reasonable level of maturity. What it now lacks most is detail (player, room, and object stats, money, ect). This release hails major gains in the areas of stability and programmability (for content developers).... read more
Just for kicks I ran a quck stress test on my latest copy of the 0.1rc2 CVS code. I'm running the mud daemon on an Athlon 1.2ghz w/ 128mb DDR ram with linux 2.4.20. The machine used to send test packets is an Athlon 750 also running linux 2.4.20. I used Ruby version 1.8.0, which I compiled myself.
Here's a quick rundown of the test setup:
2 players, each threaded, connect to the mud and start issuing commands. I see how long the mud runs before it crashes. This is a preliminary test just to get a feel for what the system can handle. It really isn't accurate to the load that would be experienced in the real world.... read more