Server monitoring

[Alex Lavrov]

Hello,

I need some information about monitoring tools.
What are the tools you recommend to use to monitor servers? File systems, applications, network etc.
I use SCM, is there any way to see the messages there?

[hpuxrox]

If your company has the money, I would suggest Openview event monitoring software.
It will tell you everything you need to know, but make sure you set up the filters becuase it will be a little to much information to start.

You can also use glance for evryday admin stuff.

[Mark Grant]

We use a combination of tools and scripts.

We believe it's important to have one central place to receive alerts of impending problems. We run Openview ITO for this and use it to alert on various issues. Things like filesystem usage monitoring are really difficult to achieve via these tools because they are so site specific so people usually end up scripting it.

We also felt that having fewer alerts that always got read was more important than lots of alerts that weren't quite as critical. Iw ould suggest you get as few alerts as makes sense as you can get awaya with.

Specific O/S commands that are useful for getting statistics are glance or gpm, vmstat, iostat and my favourite "ps".

It's a big subject and you need to decide how important monitoring is and set that against cost. I have seen systems that have massive video screens showing network stats, server stats etc etc and you could drill down in all these things. On the other hand, just keeping a glance window on your machine might be sufficient.

[Graham Cameron_1]

... and if your company has no money, I would recommend Big Brother.

www.bb4.com

-- Graham

[Marco Santerre]

Openview is the tool out there for the monitoring of your systems. You can even apply some add-ons to monitor a little bit more, like your Oracle database, and SAP applications. I'm not sure what the real name is anymore (as HP changed it from ITO, to VPO, and OVO), but I'm sure that if you tell them that you want more info about Openview to monitor your systems, they'll know what you talking about.

[Alex Lavrov]

What can u say about Big Sister?
or Big Brother?

Any good/bad comments?

[Pete Randall]

Alex,

Simply because nobody else has mentioned it, I have to bring up Glance/MeasureWare. Used in combination with some cron'd scripts to check file systems, etc., Glance can keep you abreast of what's happening. Regular observation gives you a feel for what your system *should* run like so you can tell at a "glance" when it isn't. Using MeasureWare allows you to automate this observation (to some extent - I still advocate personal familiarity).

If you don't have a lot of money, you can still do a good job of monitoring with just the system provided tools (sar, vmstat, swapinfo, etc), some scripting, and cron, but, for just a little more money, I would recommend that you have to have Glance at a minimum.


Pete

[Victor BERRIDGE]

Hi,
Big brother does a good job depending what you want to monitor...
It displays its information via a web server.
Its easy to configure...
And isnt resource greedy like patrol...


All the best
Victor

[Jon Mattatall]

We currently use (among others) BigBrother and Patrol/BMC to monitor a mixed UNIX/MS environment.

BigBrother is simple and low overhead. We run our BBDISPLAY on an old P3 we had lying around. It is very portable. We currently run it on Solaris, HPUX, NT, 2K, and SCO boxes, and monitor routers/switches for connectivity as well. If you can script it, you can monitor it with BB.


Patrol/BMC is expen$ive. It can, however, gather a greater variety and depth of stats right out of the box. Virtually every Oracle stat is available (free extents, tablespace % full, etc, etc.), but it costs. With BB, we just check for Oracle processes on the box and do a quick query to be sure the instance can be reached.


It depends on who will be doing the monitoring and what you want from it. BB presents info as it appears on a terminal - it's all shell scripts. I'm comfortable with that. Patrol gives you the full flashing lights, fuzzybumpyhappypointyclicky Windows experience.

Cost is likely to be the big factor.

[Alex Lavrov]

We currently have Tivoli .... but I'm not extra happy with it.
I think it's too big and complicated for our needs, so I'm looking for other alternatives.
BB I think too simple (we jhave a big unix environment).
Maybe I'll try OpenView ...

[Mark Grant · Accepted solution]

I think you might find Openview as big and bloated as Tivoli unfortunately.

[doug mielke]

The listed tools are great, but cost $$.

If budget is a concern, there is much that can be done with sar, vmstat, lanadmin.

Sar can save history, can run on configured schedules, is very well documented, and for the areas covers, is surprisingly complete.

You could consider it Unix's 2 cents worth.

[Brian Lee_4]

If you don't have a budget, you can use "Big Brother" to monitor your systems.
If you have some budget available, you had better Tivoli products.

We are running Tivoli in out data center.
It works great for us. It doesn't only monitor but also troubleshoot system problems if you are good scripter.

For example, you create the script increasing file system when one of file systems hits 98%. When the event pops up saying the file system reaches 98%, it runs the script automatically.
Thus, you don't need to touch your keyboard.

You need at least three Tivoli modules.
Tivoli Framework
Tivoli Distributed Monitoring
Tivoli Enterprise Console

Refer to the following web site.

http://www-3.ibm.com/software/tivoli/solutions/pa/monitoring/