UNIX for Dummies Questions & Answers

Thanks,i will check this tool ...
There is very strange thing, apparently that nobody don't need to monitoring unix system in high degree...

Are you kidding?

Unix monitoring tools are far more advanced and powerful than anything available for Windows. It's just that your Windows-centered requirements are not met since Unix apps won't be written to exactly copy what you see in Windows.

Where I work now we have over 4000 Unix servers. All of them are monitored at all times and we have set up alerting to send emails, pages, or both when certain criteria are reached. We can customize the criteria based on which categories a server belongs to. For example, the same error might cause a production server to send a page, but a development box to just send email. We can control who gets the notifications the same way so each group of admins and users gets alerted for only the servers they care about. We can generate reports per server or per group of servers containing all the data you mentioned plus lots more. I would certainly consider that a "high degree" of monitoring.

As System Shock mentioned, Unix has the commands built in to do exactly what you want, if you drop the extreme restrictions. If you really need to buy a commercial product you could look at Nagios with purchased support, Tivoli, HP Openview, several products from Computer Associates, Sysedge from Concord Software, Splunk, or many other SNMP based monitoring products. All of them have sophisticated reporting features and a gui dashboard, but they won't be exactly the same as your windows monitoring apps of course.

The bottom line is if you want powerful, customizable Unix monitoring that can give you exactly the information you want you have many choices that can do it. If you expect it to also spoon-feed you reports in the specific canned format of your choice or show you a gui exactly like your windows boxes have you may be out of luck.

Thanks...

" Nagios with purchased support, Tivoli, HP Openview, several products from Computer Associates, Sysedge from Concord Software, Splunk"

In Which tool your company is using?

Are you sure that all of this tool can give me page/sec pagefault/sec ,handles and other counters per process?

I download SiteScope of Mercury ,and for processes I have only cpu ,memory and threads ...

The company I work for now uses Nagios, Tivoli, and Splunk. At my previous job we used Sysedge in my department and I believe Openview in others.

I'm sure any of those tools could do what you are asking with some effort on your part even if they can't straight out of the box. They all use SNMP for handling events and communication. You can write scripts defining exactly what you want monitored, then let the tool take care of the alerting, logging, etc. using SNMP. Of course, all come with plenty of pre-built monitors for things many people want to see. But if you want to monitor something they don't have built-in you can always write your own script. For example, with Sysedge at my previous job we wrote a script that listed exact text strings we wanted to monitor in our /var/adm/messages file on Solaris to catch errors specific apps we used generated even though Sysedge couldn't monitor those apps out of the box. You could do the same thing, or hire a consultant or pay the company to write them for you as part of the implementation.

As mentioned before, you could also do it yourself. At my current job we also have a home-grown tool that generates graphs of just about any performance parameter you can see with standard Unix commands. It collects data on each server then runs a lightweight http server so you can use a web browser to see tables or generate graphs. Specific to your needs, it can do pages in and out per second, pagefaults per second and similar data. I don't think it can do it per process though. Although again, since we wrote it ourselves if we needed that data we could adjust the data collection scripts to implement that.

I guess that's just more details in support of my original point. The tools exist to get what you want. But it will probably require modifying a commercial app or writing pieces on your own to get some of the specific things you requested.

As a minor digression - that's part of the difference in philosophy between Unix and Microsoft. Unix tends to give you lightweight but powerful tools. You can do just about anything with them, but you need to do some work to get what you want. Microsoft on the other hand gives you more "complete" and finished tools, but if they don't do exactly what you want them to you're probably out of luck.

Thank you on detailed explanation...
I am going to check "Tivoli" tomorrow ,but from your experience which tool (Nagios, Tivoli, and Splunk) can give me remote access that is I want to see all data of Unix monitoring on my windows 2003,and which tool has a more counters for Unix ?

Preferable less employment of personally unix using (only installation Agent or something) because Unix is my weak place

I would expect Tivoli would have the most built-in options as it is a well-established product which has been around for years and is used by many huge companies.

Nagios is an open-source product which is quite robust and mature, but not as much as Tivoli. However, if you're willing to go the open source route you might find scripts and so on on the web to use with Nagios to implement what you want. From that point of view Nagios might have more available overall if you consider community support plus the official offerings.

I don't know as much about Splunk as we're just implementing it here. It can do a lot of stuff, but I think you need to script a lot of it yourself. I would expect it doesn't have as much built in functionality as either Nagios or Tivoli.

I know you could use a Windows box as your console for Nagios and Tivoli. I suspect it for Splunk too but I'm not positive.

Thanks.

http://www14.software.ibm.com/webapp...d.jsp?b=Tivoli

What I need to download in order to monitor unix system recourses on windows?