Premptive Monitoring of RHEL Servers

[Maaz]

We are going to deploy Application Servers, File Servers(NFS), Mail Server(Postfix), DNS Servers(bind)and Database Servers on RHEL 4 Servers.

My responsibility is to check the servers health/status to keep them optimize using preemptive maintenance/monitoring

Please recommend me the CHECKLIST/STEPs to monitor the Servers for preemptive monitoring.

like e.g, I mean please recommend me that
1) what to look ?
2) where to look ?
3) how to plan the capacity planning
4) what logs to check
etc for Preemptive Monitoring of the servers.

Please share the steps/checklist for the preemptive monitoring

One more thing that none of the rhel 4 servers will be configured to get updates directly from Redhat Network, because we want to apply updates/patches in a controlled manner. So does the Redhat Sends the update/patches media to their subscribers. So that first we check the updates by applying them to testing server, and if the updates are perfect install(manually) them to the production server ?

Regards

[Ivan Ferreira · Accepted solution]

1) what to look ?

Considering the type of server, you need a monitoring tool like nagios and look for:

- File system space
- Server status (is it up?)
- Services response (is the service answering requests?)

2) where to look ?

Nagios statistics

3) how to plan the capacity planning

Use the staticstics recorded by nagios to plan future grow in CPU/Memory/Disk

4) what logs to check
etc for Preemptive Monitoring of the servers.

Use a monitoring software like Proliant Support Pack and configure it for email notifications.

>>> we want to apply updates/patches in a controlled manner.

You need to configure the Red Hat Network client to "only download" the updates.

>>> So does the Redhat Sends the update/patches media to their subscribers.

No. Updates must be aplied from RHN.

>>> So that first we check the updates by applying them to testing server, and if the updates are perfect install(manually) them to the production server ?

This is why you need to configure the client as "download only".

[~sesh]

If you are talking about HP servers predominantly, then you can use HP Systems Insight Manager which gives you a graphical representation of a server's status, any device failure, server down/up etc.

So your question 1 & 2 are answered there.

You also have ProLiant Essential Performance Management Pack, which lets you monitor the performance and plan for any capacity improvements as necessary. (for question 3).

Various logs to check, from basic /var/log/messages to cfg2html, ADU, Insight Diagnostic logs etc.

Hope this helps.

[Maaz]

Thanks Ivan and adric for help.

only the Database and App server will be running on HP Proliant machines thats why Proliant Support Pack, HP Systems Insight Manager, ProLiant Essential Performance Management Pack are not the complete Solution to me.

Postfix, Fileserver, Bind, and Web-Proxy servers will be on low-price comodity hware.

nagios is a solution that covers every machine in my case.

in my case forget about the tools, please let me know all those log files, and default utilities that ships with rhel 4 servers for the Proactive/Preemptive Server monitoring (mail, dns, file, database servers)

Thanks for informing me the various tools, but I am more interested in the CHECKLIST a person must/should have, to monitor the Servers Pro-actively/preemptively. i.e in other words one can use any utility/tool to monitor, but that tools must check a set of things or number of things

In other words, I mean i need a checklist/STANDERD/list type thing, so that any techy can use use his choice of utility/tool(commercial or free)... no problem... but that utility must covers/checks those things(staderd/list)

[dirk dierickx]

nice OSS tools to monitor a linux servers are:

collectd - http://collectd.org/
mon - http://mon.wiki.kernel.org/index.php/Main_Page

[Ivan Ferreira]

From the performance point of view, there are a excelent tool that you should install. The "by default installed" tools are not as good as this one.

http://collectl.sourceforge.net/

[Mark Seger]

Ivan - thanks for the recommendation for collectl as I am the author of it and always looking for happy users. Be sure to periodically check back at sourceforge because I'm always updating it. In fact I recently released a version that knows how to display process i/o stats for kernels that report them (I think 2.6.22 and above) as well as the new slab allocator stats in the newer kernels as well.
-mark