[ale] server question

[Transam@cavu.com]

The problem is that we are running out of REAL RAM (not VM or disk).  Each
of the hundreds of mysqld processes uses up to 6 MB or more of RSS (actively
used memory) and 15 MB or more of VM after the system has been running for
3-4 hours.  When first booted each of the syslogd processes uses a smaller
amount.

There seems to be a one-to-one correspondance between apache and mysqld
processes so we assume that each apache user spawns a syslogd process.
I suspect that the bug is that there is a memory leak in mysqld.  The
problem has been getting worse over time but that may be due to increased
usage.  (A client with a rapidly growing business, whoopie!)

The system and software is ancient.  We aren't the ones who maintain the
software it, just reboot it at 3 am or, ahem, shortly before the premier
showing of Pearl Harbor.  Grrrr.

Bob Toxen
transam at cavu.com                       [Bob's ALE Bulk email]
bob at cavu.com

Sage (Indira) wrote:

> Hi Leonard! *wave* I guess you haven't killed anyone at ATT yet? Did
> anyone ever come out to repair your lines? The guy who installed ours
> said that the ATT repair dept. was understaffed with clueless boneheads.

> Free disk space? none. It looks like mysql may be what's eating it up,
> but, like I said, I could be wrong.

> Ummm...this was rather stupid of me, but I don't recall what the kernal
> panic message was...I'll see if Bob remembers. As I recall, though, it
> didn't offer a whole lot in the way of information. However, I remember
> we checked syslog, and it offered no clues, other than the system ran
> out of real memory (not virtual memory).

> The only message that has been consistent is "Can't find a free
> page". Sound familiar to anyone? What I'm finding really weird is the
> periods between downtimes decreasing exponentially. When I get to work,
> I'll try logging on to that piece of crap, and see if I can find anything
> helpful.

> Thanks!

> -Indira

> -Sage


> On Wed, 13 Jun 2001 12:49:04 -0400 Leonard Thornton <Leonard at Intelis-inc.net> wrote:

> The first thing I would check is the free disk space.

> Secondly, what are your panic messages telling you?  What about 
> syslog?  Enquiring minds want to know....

> At 11:34 AM 6/13/2001 -0400, you wrote:
> >Heyla, all!
> >
> >We have a server doing a really odd thing. Since November, the time this 
> >stupid thing stays up without a reboot is decreasing exponentially. It 
> >started off having to reboot every four months, and uptime without a 
> >reboot is now about every 6 hours. Last week, we finally had to set a 
> >crontab entry to automatically reboot the stupid thing every 24 hours, 
> >then every 12, then every 8, and now, every four. (I got a bit ill having 
> >to run downtown 4 days in a row) I predict the damn thing won't live 
> >through the weekend.
> >
> >It seems to have an awful lot of mysql processes running when it's up. The 
> >mysql version is 3.21.33b. Here lately, we've also been getting kernal 
> >panic messages. The kernal version is 2.0.36. (Am I spelling kernal 
> >right?) Yes, I know it's as old as hell, should have been upgraded, yadda, 
> >yadda, yadda. That would be the client's call, and their option was just 
> >to replace the whole box, but they can't seem to locate the people who did 
> >the old machine, and we're kind of running out of time.
> >
> >Question is; has anyone come across a problem like this, and was there a 
> >solution? It looks like it may be possible that the mysql processes aren't 
> >getting killed after users log out, but I may be wrong.
> >
> >Any suggestions would be appreciated.
> >
> >-Sage

> The difficult while you wait.....the impossible overnight.

> Leonard Thornton
> Intelis, Inc.
> 5960 Crooked Creek Rd
> Suite 30
> Norcross, GA  30092

> Office: 770.825.0032
> Fax:            770.825.0028
> Cellular:       404.583.5402
> Pager:          888.785.9188
> Email:          Leonard at Intelis-Inc.net
--
To unsubscribe: mail majordomo at ale.org with "unsubscribe ale" in message body.