Hello folks,

just 3 tips to prevent your server from going down.

• ALWAYS keep your server patched up to date.
  - You can find official Sun/Cobalt patches here:
  http://sunsolve.sun.com/pub-cgi/show.pl?ta...g&nav=patchpage
  - Unofficial Sun Cobalt patches (since Sun is sometimes somewhat slow) here:
  http://www.raqtweak.com/free.php
  IMPORTANT: If you do not keep your server up to date, you are opening all doors for HACKERS!
  Beware, I've seen it happen QUITE OFTEN!!
  (Alternatively you may want to look into firewall solutions, etc...)
• Note regarding KERNEL patches:
  DO NOT install them through the GUI using the URL option.
  They will go corrupt and they will crash your server (well, the chances are VERY high).
  Instead, login to your server using SSH (latest version can be downloaded free of charge at http://www.raqtweak.com/free.php - telnet is a security leak risk!),
  "su -" to root,
  then "cd /home/packages" and then
  "wget http://"
  Go to the GUI, and select "Loaded package".
  This WILL work!
• Keep your root parition fairly "emtpy".
  Login to your server (using SSH), and go "su -", then "df".
  The % of any partition should NEVER go over 80% just to be safe.

Just a few tips on things I've seen going wrong many times...
Good luck with it!

Good Tips, ill keep them in mind for future updates.

In the past I have installed the kernel updates via the GUI by URL, not had any problems with that method thank god, I must be very lucky... I will do it your way in future, best not to invite trouble

regarding space on my root partition, I seem to be running a little low... 79% used
I dont think there is a whole lot I can do about that as I have already had your root partition fix installed... Any ideas what I can do to free a little space?

Indeed very lucky. I have done it myself too, and it worked sometimes... But I have seen it go wrong too often, hence the warning.



Hmm... Contact me... I will have a look...

Dear Raqtweak, I followed your instructions and installed successfully.
I have seen that the package is still in packages directory, should I delete or move it maybe? Is there a particular command?
I also tried the "df" and had these results:

/dev/hda1 743466 582976 160490 79% /
/dev/hda3 198601 19578 179023 10% /var
/dev/hda4 17671528 1831828 15839700 11% /home

Do you think that I should do something?
(As you may notice, I'm not very expert!)
Thanks!

You can delete the PKG once installed, no problem.

the 79% is what we are looking at here...
79% is still "ok"... But could be better...
if it goes over 80% I stronly recommend you would have someone look at it

all-cards - it looks ok for me too

If it move more close to 100 % you can
out store stuff to your /home/ and set
a link to it (this will tell programs to follow
it to the new location).

Thanks OwNeR!
I will check from time to time, and if it goes up, I'll ask for help for sure, as I'm not expert even I would like to learn. I understand what you said about moving some stuff and setting a link, but I don't know what and how :confused:
This is really a great community!

Where it says 'Reboot Required: Yes'

is it talking about rebooting apache

/etc/rc.d/init.d/httpd restart

or

/sbin/shutdown -r now

If it's the shut down code, how do you turn it back on, or does it do it automatically. I'm going to do all 38 patches and do one at a time, rebooting every time I add one that requires it. Will this also help lower the server load??? About daily I'm geting atleast one of those E-Mails...

*****

Over the past fifteen minutes, the CPU has been heavily loaded.

1 minute load average: 44.21
5 minute load average: 28.73
15 minute load average: 14.03

*****

Hi Jesse,

what they mean with reboot is:

shutdown -r now.

-r is REBOOT, which reboots the unit, like a fast power off and BACK ON.
-h is HALT, which turns the power OFF (and DOESN'T turn it back on)

As for the CPU load, the patches probably won't fix that... Do you have a forum running on that thing?

I've had a vBulletin board on it for two years. I just went four months with no crash, though it looks like the next crash will be in less than four months since those E-Mails are here.

Filesystem 1k-blocks Used Available Use% Mounted on
/dev/hda1 743466 575662 167804 77% /
/dev/hda3 198601 108702 89899 55% /var
/dev/hda4 17671528 3027701 14643827 17% /home

And is it possible to do the shutdown -r now command too many times? When I first got the server, I did it a few times as it looked like the server was going to crash, then the next time it did a crash, re-booting didn't work so they had to take it to there shop or something to fix it, so it was down for a few days. Ever since then I havn't dared do the 'shutdown -r now' command.

run a "top",
do you see anything that uses a lot of CPU?

*edits last post*

load average: 0.66, 0.94, 1.15
CPU states: 23.0% user, 9.3% system, 1.3% nice, 66.2% idle

It looks like it get's a spike about once or twice a day.

will

cd /etc
pico -w crontab

01 * * * * root /usr/sbin/logrotate -f /etc/logrotate.conf

or delete the logs help lower the server load?

Rebooting servers should really break anything...

As for the logs... Do you want them? You can also turn them off...

Nope, I don't need them. I would love to turn them off. I can make my own logs using perl if I ever need them.

chmod 000 /etc/cron.daily/log*

That should turn it off... let me know if it worked for you...

Thanks. I'll know with in 24 hours. *deletes logs*

Hi Raqtweak,
I sent you an email through your site, but I try to contact you here too, I need urgent assistance please.
Thanks,
Andrea

[root /]# df
Filesystem 1k-blocks Used Available Use% Mounted on
/dev/md1 743371 617834 125537 84% /
/dev/md3 198275 21562 176713 11% /var
/dev/md4 26918208 1417435 25500773 6% /home

4% too much, what can I do to fix this?

Hi,

My RAQ4i seems to be beyond redemption by these standards - root partition is 90% used! What do I do to free up space?

Plus I just checked "top -s" now and it shows,

=====
128 processes: 125 sleeping, 3 running, 0 zombie, 0 stopped
CPU states: 11.7% user, 72.8% system, 15.4% nice, 0.0% idle
...
=====

and

=====
790 root 20 0 6156 6156 3256 R 0 65.3 2.3 32:09 caspeng
13214 postgres 18 5 2924 2924 2420 R N 0 36.8 1.1 3:20 postmaster
=====

So the asp engine and the postgres postmaster between them are using a lot of cpu resources. Any tips on how that could be prevented?

On another note the webalizer daily job also seems to be taking up a huge amount of resources - what can I do to permenantly disable logs and stop webalizer from running?

raqtweak - the "chmod 000 /etc/cron.daily/log*" command seems to disable the logrotate program alone. Does this stop generation of new log files?

BTW in.qpopper seems to be using up a lot of CPU resources (especially when 5 or 6 concurrent processes are live the load average shoots beyond 5 or even 10 at times!). Does anyone know why this happens and are there any solutions?

Thanks,
-- Santosh

Hi Guys,

My webserver keeps crashing and I am worried as the Cobalt OS is getting old.

I am thinking about using this new Strongbolt software from www.osoffice.co.uk has anyone else used it? Is it any good?

I noticed that it is like Cobalt OS, but has Apache 2. Maybe this is more stable?

Pango

Actually, have you seen:

http://www.rackstar.net/

I've had a vBulletin board on it for two years. I just went four months with no crash, though it looks like the next crash will be in less than four months since those E-Mails are here.