CLX archive - 1998: RE: Some questions...

RE: Some questions...

=?iso-8859-1?Q?Thomas_M=F8rch?= (uhdpxeg@roskapostia.tunk.net)
Mon, 26 Oct 1998 18:04:03 +0100

>----------
>From: Bill Strickland[SMTP:vtigzd@norlight.com]
>Sent: 26. oktober 1998 17:16
>To: clx listserv
>Subject: Some questions...
>
>During the CQWW test this past weekend, I had a Saturday crash and
>a Sunday crash. There were only 3 users (only 2 via RF connects).
>Simple commands took FOREVER to execute (like sh/us, set/nodx, etc.)
>
>1. How do I try to fix this? More memory? Crashes usually report
>that the Shared Memory Manager is "not OK". Will 32 Meg help this?

I use Redhat 4.2 with kernel 2.0.34, on a 486dx2 66 with 48Mb ram,
havent noticed any delays at all... Right now it has been up for 23 days
(What
is the curent record???)...
>
>2. Go to Pentium? I'm not anxious to spend a lot to upgrade, but this
>may be a necessity for stability. What speed are YOU running?
>
>3. How often SHOULD vacuum be run? Should it be turned off during
>contests? Set to run more often during contests? It seems to grab
>a lot of resource time when it runs.

I think i run it once every hour (Can't remember it, and i don't have it
nearby)
>
>4. Did your clx node stay up all weekend? How many users did you have?
>What is your hardware configuration? I may have something basic that
>is not configured correctly making the system lock up.

It seems like it did stay up the entire weekend, don't know how many
users
i had though..

Also i run the watchdog every minute, so it should catch a crash
quickly, if
there should be one, and restart the cluster...

- Thomas - OZ2AFO - Sysop on OZ2BOQ dxc/bbs/node system...