MOO-cows Mailing List Archive
Re: Some Technical Questions
Date: Sat, 26 Aug 1995 16:10:47 PDT
From: email@example.com (Matt Pauker)
Content-Type: text/plain; charset="us-ascii"
At 2:45 PM 8/26/95, Colin Moock wrote:
>Since starting VUW MOO, an academic MOO, a few months ago,
>I have gained a lot of experience with the Lambda core and server, but I
>must admit that I am still only partially familiar with the UNIX server
>on which VUW runs. Because I am not the sysadmin of the server, I don't
>get to experiment much with figuring things out, so I have a few
>important questions hanging around my head that I'd really like to get
>answers to. I hope someone will be able to help me out with these...
>1) The server occassionally crashes (1-3 times a month). Is this normal
>for LambdaMOO-1.7.8p4? Any ideas why it crashes?
>I have checked the logfile for answers, but it is always silent (that it,
>it makes no mention whatsoever of crashing). ./restarting the MOO always
>works fine too. I asked the sysadmin if anything on the host machine
>would bring the MOO down, and he said it *may* have run into a memory
>limit which shut it down (but he doubts it because he thinks it is quite
>high, and figures it would happen all the time if this were the case).
>Which leads me to question 2:
>2) How does the sysadmin check how much memory the MOO is using? What
>are the kinds of memory he can check? (I assume he knows, but I'd like to
>know too.) Can I check them or must the root check? I'm running the MOO
>on a UNIX DEC ALPHA model 2100 running OSF/1 ver. 3.2a
>3) What from within the MOO is the greatest concern for keeping lag time
>down? (should there be, or is there, a "Everything you want to know
>about minimizing MOO lag" FAQ?)
>4) How does the amount of MOO users affect lag and memory usage?
>(perhaps there's a general estimation formula, like 1 user= x seconds
>lag, y KB ram etc?) Dont worry about db disk space...I have plenty.
>I appreciate all of your input, especially those of you running academic
>MOOs who will hopefully empathize with my situation. Currently, there are
>only ever about 5 users on the MOO, but in two weeks, it opens to the
>students taking the course for which the MOO was constructed, at which
>time I expect an average of 25-50 users. (which is why question 4
>concerns me...I don't want the sysadmin to undergo a shock at a massive
>increase of memory usage, nor do i want the MOO to wallow in 10 seconds of
>If I receive enough info., I may compile a short FAQ on memory usage and
>LAG for newbie arch-wizzes.
>As ever, thank you for your support, without which I'm sure I wouldnt be
>where I am now...
>Waterloo, Ontario, Canada
I'll answer as many of these as I can...
1) It shouldn't normally crash like that. One possible cause of it
shutting down unexpectedly would be someone rebooting the machine on which
it is running. You might want to ask the site admin if the machine was
rebooted during the time the MOO was down.
2) The easiest way to check memory usage is with the 'ps' command. The
arguments to use vary from OS to OS. I'm not sure what arguments to use
with that OS, but with Linux, I normally use -aux.
3) There are a lot of causes of lag. The speed of the machine makes a big
difference, as well as how much other usage the machine gets. A slow
connection (Netlag) can be very noticable as well.
>From within the MOO, causes of lag are forked/suspended verbs, many player
connections, and calls to open_network_connection().
4) There really is no formula, since there are so many variables. You'll
just have to experiment with your MOO and see how many users you can take.
Hope this at least answered some of your questions...
Subject Index |