simple page request check?

[polx]

Hello HypericHQ community,

I have successfully installed Hyperic HQ on my MacOSX 10.6. It auto-detects Apache and Tomcat well and can show me some statistics.

However, I would like to know if it is possible to add an extra monitoring-source: the GET query of a web-page of which one could plot the load time and, more importantly, check for an expected substring and trigger alerts if this is missing.

I don't see a way to do that currently.

At the Apache resource, I can only add a mod_jk_worker service...

Do I need to fetch/write/install new plugins? If yes, which?

thanks in advance.

Paul

[zippyz]

You could write your own curl tool using the script plugin, which will allow you to generate your own custom plugin that can retrieve data from script tools. http://support.hyperic.com/display/DOC/Script+Plugin

I wish Hyperic had a way to do this natively, I'm pretty sure Zenoss has this kind of functionality built in (It uses the Bottle Python Web framework I believe).

[polx]

Le 2 juin 2012 à 01:52, zippyz a écrit :

You could write your own curl tool using the script plugin, which will allow you to generate your own custom plugin that can retrieve data from script tools.http://support.hyperic.com/display/DOC/Script+Plugin

Thank you Zippyz,

Yes, I think that looks like a good idea however the script plugin has a single tutorial and that tutorial is already far too big! Is there a simpler tutorial??

Maybe a tutorial where a plugin would be written in groovy would also be extremely simple, or even in java. All of that I could make, but not so complex please for starters!

With such a complexity around, I am more and more inclined to to switch to a hand-based method as I have thus far whereas, precisely, the interest of HypericHQ, or another integrated system, is to have a holistic view with ready-made tools for the management of elementary things such as memory or process load.

It would be a pity because HypericHQ has some impressive facets, including its shiny startup scripts.

thanks in advance

Paul

[hqpso]

There is a builtin platform service that performs this function out of the box.

See details here: http://support.hyperic.com/display/EVO/HTTP+Platform+Service

The HTTP Platform Service will also do pattern matching as well as reponse time and response code metric collection.

[polx]

There is a builtin platform service that performs this function out of the box.

See details here: http://support.hyperic.com/display/EVO/HTTPPlatformService

The HTTP Platform Service will also do pattern matching as well as reponse time and response code metric collection

HQpso,

Thank you, that just saved it!!

I wonder if this should not be somewhat more visible... there's a huge wealth of platform services (e.g. even IMAP) but one has to know that one has to search them at that level, and at no other (e.g. not at the Apache level, nor at the root level, though both are potential places).

Indeed, a predefined service is what it should: it carries altogether the zillion metrics I would never programme for but might, one day, become interesting (response time, all tcp-connections stats, ...). Typically what one expects of a packed monitoring tool (as opposed to Graphite or my own tools, for example).

Do I understand correctly that this service shows unavailable if the regexp is not matching?

If yes, I'll adjust the doc for that because it is not 100% clear.

Based on that we can say the alert "this resource is broken" but would there be a readable place to diagnose the finding that the status is ok but not the regexp?

Thank you really much, finally I'm seeing the end of this dark tunnel!

paul

[hqpso]

paul,

If you configure an HTTP Platform Service and use the pattern matching, when the page responds but the string is not found, Availability will turn Yellow and show 50% availability. If you create an Alert for Availability less than 100% an Alert will be triggered. If the page does not respond in the configured 'sotimeout', Availability for the Resource will be 0% and Red.

If you configure log tracking to Debug or Info you will get an Event that's states the response code, however this is already shown in the metric indicator 'Response Code".

Using multi-criteria Alert Definitions, you could create a variety of conditions..

[staceyeschneide]

Thanks, hqpso!

[zippyz]

I don't think HTTP Platform service will work. It only lists protocol level metrics, rather than being able to extract specific metrics from the page. For example, if I had a page with the following JMX metrics available:

ProductsCache: 880

ServiceCache: 91

OrderIDCache: 991

Zenoss seems to be able to do this natively.
Although the script plugin is very powerful, something like Zenoss which includes Bottle or Mechanize will do a much cleaner job of extracting metrics from the page.

[admin]

zippyz,

You're right, for the example you gave the HTTP Platform Service would not be suitable.

For the question that was asked by paul, it will do exactly what he wants.

For your eample, a Script Pulgin would work if you wanted write your own. Better yet, if you have JMX access to the platform housing the information you are interested in, (assumed since you implied the metrics shown are JMX MBeans) it would make more sense to use the builtin Hyperic Sun JVM Server to gather the metrics.

This is available out of the box.

Hope that helps.

Cheers.

[admin]

zippyz,

You're right, for the example you gave the HTTP Platform Service would not be suitable.

For the question that was asked by paul, it will do exactly what he wants.

For your eample, a Script Pulgin would work if you wanted write your own. Better yet, if you have JMX access to the platform housing the information you are interested in, (assumed since you implied the metrics shown are JMX MBeans) it would make more sense to use the builtin Hyperic Sun JVM Server to gather the metrics.

This is available out of the box.

Hope that helps.

Cheers.

[hqpso]

zippyz,

You're right, for the example you gave the HTTP Platform Service would not be suitable.

For the question that was asked by paul, it will do exactly what he wants.

For your eample, a Script Pulgin would work if you wanted write your own. Better yet, if you have JMX access to the platform housing the information you are interested in, (assumed since you implied the metrics shown are JMX MBeans) it would make more sense to use the builtin Hyperic Sun JVM Server to gather the metrics.

This is available out of the box.

Hope that helps.

Cheers.

[zippyz]

Very well. Thanks for the replies. In the case of the Zenoss community, I essentially didn't receive one. Hyperic 1, Zenoss 0.