System Administration

[simon_164]

What are some steps to monitor the system daily?
what log files to look for?
what steps to do?
If anybody can list the things he does each day to monitor his unix system it will be wonderfull.

Regards,

[Peter Godron · Accepted solution]

Simon,
log onto the server and check /var/adm/syslog/syslog.log for errors
run dmesg
review backups were ok
check the filesystems for space
check, with top, for run-away processes
Speak to users, to find out if anything unusual is happening
.
.
.
.
or install glance

[James George_1]

Hi

As peter Pomited out :

check your /var/adm/syslog/syslog file for errors.

# bdf ( for checiking the File system size)

# dmesg ( for any kind os system errors)

#top 10 ( to see the CPU utilization / process status)


# check your ignite backup status

Rgds / James

[sachit patil]

hi try this
1> #top
2> #tail -20 /var/adm/syslog
3> sar -u 1 5
4> iostat
5> vmstat
6> bdf
7> more rc.log
8> lanscan

[Rita C Workman]

Simon,

What others have said will be helpful, but not nearly enough.
There are many logfiles to be checked and cleaned. How is up to you and syslog is but one of them.
What software do you have installed ?
...backups logs; third party application logs; bad logins; file system full; performance peaks (CPU,Disk,memory)etc...

And a good Administrator will set up something to automate monitoring of the scripts and send alerts. Personally I send certain messages to one central server, have that logfile checked regularly & email issues to my pager & email (for me & other admin). Nobody can be watching everything all the time...
.......automate,automate,automate !

Do some searches here as this subject as been discussed...you can brew your own, or use some software to help you.

Rgrds,
Rita

[simon_164]

Thanks for your replies,

Well i have SAP installed on the servers, two of them are in a MC-SG cluster. Other than that no special software is there.

I am not familiar with creating scripts to monitor log files, however i will start to dig in here and in the books to know how to do that although having a server dedicated for the receival of these logs is not Financed in my company.

I will start to see all that. Glance Plus is not to be installed since it is not free.
I am using ISEE for the hardware problems.

[Sundar_7]

I would suggest you familiarize yourself with what seems to be the important part of your infrastructure - the MC-SG cluster.

Once you get the hang of it, you can concentrate on the bells and whistles.

Study the package control scripts - how the applications are stopped or started. What are the services configured? How many primary/lan cards does your cluster have ? Is it set to AUTOstart or for automatic failover ? How many heartbeat networks configured ?

Familiarizing with the cluster setup can save you very valuable troubleshooting time when something goes wrong. Ofcourse not many monitoring tools can impress your boss as much as putting-off fire in your cluster as quickly as possible :-) Just my 2 cents.

Good luck.

Sundar.

[M. Russell Milliron]

I also use a free tool called LogWatch:
http://www2.logwatch.org:81/
It took me a while to get the config the way I want it, but it sweeps the system log daily and produces a summary.