system checks

[Ravinder Singh Gill]

What regular checks should I do on servers ie on a daily basis.

[Steven E. Protter]

Wide open question.

1) Quick look through /var/adm/syslog.log for strange stuff.
2) dmesg for hardware issues.
3) bdf for filesysems getting full.

I recommend you write a script to scan these results and send you a daily email report. I had over 100 tools for various tasks in this area. Its based on what your server does.

SEP

[Pete Randall]

You need to check your file systems (bdf), your log files for unusual events (syslog, mail log, diagnostics, last, lastb), your hardware itself for warning lights, etc.

That's a pretty simplistic list, but it's a start.


Pete

[Pat Lieberg]

If your server is a DB or application server, you should work with whomever supports the DB or app and find out if there are specific things they would like or need monitored (e.g. filesystems, memory, daemons). Sometimes, there is no one but you and in that case, you can check with the company that made the application.

[Kent Ostby]

uptime and check your "load average".

Check root email for messages from cron about possible failures.

bdf to check disk usage

look through dmesg and syslog

The other way to go is to run EMS and have it monitor a lot of your resources for you and send you an email if something is out of line.

Best regards,

Oz

[MarkSyder]

/var/adm/btmp
/var/adm/sulog
/var/adm/wtmp
/var/adm/cron/log
/var/adm/lp/log
/var/adm/syslog/syslog.log
/var/adm/syslog/mail.log

Check they don't get too large (trim in SAM) and check the contents of btmp and syslog.

Mark Syder (like the drink but spelt different)

[Ravinder Singh Gill]

How would I know if the disks were failing/about to fail, before they actually failed? What should I be looking for and where?

[Pedro Cirne]

Hi,

HW is monitored by EMS, usually a mail is sent to root with alamars and they are also registered in /var/adm/syslog/syslog.log

You may configure EMS using SAM.

Enjoy :)

Pedro

[Torsten.]

hardware malfunction will be recognized by the diagnostic and event monitoring system. check, if root got mails from ems.

[Geoff Wild]

Install an utility like cfg2html:

http://come.to/cfg2html

or sysinfo (attached).

Run from cron:

# Run the sysinfo script
0 6 1 * * /usr/local/sysinfo/sysinfo -a -b -o /usr/tmp/`hostname`.sysinfo >/tmp/sysinfo.cron 2>&1
#
#Run the cfg2html scripts
35 6 1 * * cd /opt/cfg2html ; /opt/cfg2html/cfg2html_hpux.sh >/dev/null 2>&1
#



Rgds...Geoff

[Rick Garland]

What kind of environment are you working with?
Do you have Oracle? What other apps?

Check the /var/adm/syslog/syslog.log, do the dmesg command, check your filesystems to make sure space is OK. Any clusters in use? Check using the command 'cmviewcl -v'.

The utility cfg2html is good for providing a lot of this info.

Do you have EMS setup? This can keep tabs for you and alert you if something is not right.

[John E.Ophious]

G'day Mate,

All sorts of stuff!

File systems (bdf/df -k), mount points, syslog, any other logs for specific programs, backup logs... the list could go on and on.

If you have a significant amount of systems, developing a monitoring tool can be really useful. We're using one that is a huge gui with all systems listed. We monitor if they're up or down (using ping), mount points, syslog, file system sizes, licensing information, sar info.. anything we want really. Just make a list of everything you check and put it in a monitoring tool.. they're pretty easily available!

Take it easy partner,

John E. Ophious

[Ravinder Singh Gill]

The server is a Unix server on which many processes run and some databases are housed. However the database teams look after their own stuff. My team is simply responsible for the system in general. We do have NSM monitoring etc however I have been advised that I should be performing regular checks of my own ie checking syslog and the disks etc so that we can catch any faults before they may arise or before the alerts are raised. I know I can check syslog and bdf etc. I will not be allowed to implement any further tools as we already have monitoring in place. I was just wondering in general when checking the system myself where I should be looking ie apart from syslog and what I should be looking for. For example I was told by doing checks we may notice a disk is faulty/about to fail before it actually fails and thus prevent an incident. How would I check for this?

Thanks for your help guys

[Gavin Clarke]

dmesg loops so if you have a persistant problem it might overwrite other messages, to avoid missing messages put an entry like this in your crontab.

0,10,20,30,40,50 * * * * /usr/sbin/dmesg - >> /var/adm/messages

That way you'll have all your dmesg messages there as well.