Re: [cobalt-users] web.cache VERRY big

["Richard Green" <rgreen@xxxxxxxxxxxxx>]

Stephen,
I have 8 identically configured raq4's. All are up-to-date with patches and
all have the 'normal' extras from pkg.nl.cobalt.com. mySQL, webalizer etc.

Heres what I know about the web.cache files:
No matter what size it is, it contains 5 columns of numbers.
The first column is a number repeated almost 100 times before it increments,
the second column counts by threes, the 3rd, 4th, and 5th columns are almost
always 1,1,0 respectively.

Heres what I have learned about web.cache.new
Its an almost exact duplicate of web.cache, just a little bigger.

As you can see my knowledge of web.cache and its usefulness hasn't gotten
any better over the last few weeks.

> Why are these files generated:(web.cache); I don't need them because I use
> webalizer. Is it not possible to disable (delete) them ????

I don't know what they are used for, I don't know how to disable them, and I
don't know where they are ran from. I don't believe web.cache or its size
has anything to do with my crashing problem. Personally I would delete
web.cache and web.cache.new if the first line of the file didn't say - CACHE
type 4 produced by analog 4.16/Unix. Do not modify or delete!
If I remember correctly the raq2 had problems with analog.dns that grew to a
huge size also. That file was safely deleted though.

As I moved domains from one server to the other, the old server stops
crashing daily, and the new started crashing daily. The old still crashes
about once a week.

What I have determined is that the server appears to run split_logs and
analog. From what I gather in the knowledge base analog is ran automatically
after split_logs.  /home/log/httpd/access is not rotated. If you follow the
logical sequence of logrotate, first ran is /etc/logrotate.d/analog which
rotates /var/cache/analog.dns, which doesn't exist. Then
/etc/logrotate.d/apache is ran, the prerotate script for split_logs is ran,
and seems to run correctly. I can't tell if apache is restarted in the
postrotate script, but even if its not, that shouldn't kill the server.
/home/log/httpd/access is not rotated daily so the server has to crash while
running the apache rotation.

Regards,
Rich




----- Original Message -----
From: "Steven Depuydt - BeNe.WS" <steven@xxxxxxx>
To: <cobalt-users@xxxxxxxxxxxxxxx>
Sent: Monday, November 26, 2001 10:07 AM
Subject: RE: [cobalt-users] web.cache VERRY big


> WHat software do you have installed.
>
> I see that you have the problem with a RAQ 4; i have a RAQ3i
> And I took the time to stay up at night watching at TOP and by 6 a clock
in
> the morning (and after a lot of coffee) when logrotate was genarating the
> web.cache files for a verry busy site all my memory was gone.
> My server does not crashes every day but I can say a least once a week.
> After a reset everyting is OK.
>
> I installed; php; mysql; webalizer form source.
> Everything works fine. When I call to cobalt they wont give me support
> because I installed those programs.
>
> ---------------
>
> Why are these files generated:(web.cache); I don't need them because I use
> webalizer. Is it not possible to disable (delete) them ????
>
>
> Regards,
>
>
> Steven
>
> -----Oorspronkelijk bericht-----
> Van: cobalt-users-admin@xxxxxxxxxxxxxxx
> [mailto:cobalt-users-admin@xxxxxxxxxxxxxxx]Namens Richard Green
> Verzonden: zaterdag 24 november 2001 0:33
> Aan: cobalt-users@xxxxxxxxxxxxxxx
> Onderwerp: Re: [cobalt-users] web.cache VERRY big
>
>
> Stephen,
> I have the same problem I have been trying to track for weeks. My server
> goes a bit farther a crashes every day running cron.daily. The web.cache
and
> web.cache.new files don't even appear to be rotated, unless they are
> supposed to be rotated outside of cron.daily. At least I can't find where
> they are supposed to be rotated. This problem has taken a back seat to the
> server crashing daily.
>
> Originally I thought the problem was with an old Raq4. It started failing
> months ago. Every error from missing site admin, postgresql not
restarting,
> etc, so I decided to replace and restore it. Since the migration utility
> doesn't work without postgresql I restored each site manually. I moved
about
> 50 sites to a server that had been running without fail for months, and
the
> same problems started appearing.
>
> >From what I can tell, my server crashes while rotating the access log. It
> appears that the logs get split, gets placed into each sites logs
directory,
> then crashes without rotating the access log. Before the problem became a
> full blown crash, the system would mail out of memory messages. I believe
> the problem is due to a busy server and a large daily access file, because
I
> removed half the sites from the old server and it no longer crashes. I no
> longer recieve messages from that server because I got tired of the
> postgresql can't be restarted message every 15 minutes. Though the old
> server doesn't crash daily anymore, ASP dies daily and the logs don't
> rotate.
>
> I've checked everywhere, including Sun Forum for the web.cache and
> web.cache.new problem. Sun doesn't respond and no one else appears to know
> what the problem is. I only have 8 raq's so I don't have enough pull to
get
> any response from Sun. I have in the past shelled out my credit card to
> receive support from Cobalt and other places, and it seems to me, every
time
> I do, I stumbled onto the resolution before my paid help does
>
> Regards,
> Rich
>
>
> _______________________________________________
> cobalt-users mailing list
> cobalt-users@xxxxxxxxxxxxxxx
> To Subscribe or Unsubscribe, please go to:
> http://list.cobalt.com/mailman/listinfo/cobalt-users
>
>
> _______________________________________________
> cobalt-users mailing list
> cobalt-users@xxxxxxxxxxxxxxx
> To Subscribe or Unsubscribe, please go to:
> http://list.cobalt.com/mailman/listinfo/cobalt-users
>