[Nagiosplug-devel] proposal for new plugin check_byhttp

Thomas Guyot-Sionnest dermoth at aei.ca
Tue Sep 23 04:13:15 CEST 2008


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 22/09/08 11:32 AM, David Diffenbaugh wrote:
> I have a need for a plugin which would have the functionality proposed 
> below. Before I start working on it does anyone have any suggestions or 
> requests? Is anyone already working on something similar?
> 
> check_byhttp will be a nagios plugin which retrieves content from a 
> specified host using http. This file will contain a status state 
> (OK,WARN,CRITICAL) and a status message. The plugin will return the state 
> and status message that is listed in the retrieved content. It also will 
> check the last modified time of the document, as reported by the 
> Last-Modified field, to ensure the status is current.  If the file is 
> current it will report the status. If the file is not current it will 
> return a state and status message reporting that the document is not up to 
> date based on time thresh-hold arguments passed to the plugin.

I've been thinking about that lately (I probably even posted about it in
this mailing list...) - you should be able to define the string match
that cause ok, warn and crit, and possibly print the text returned.

It should probably also be able to do SOAP and/or XMLRPC requests (I'm
not very knowledgeful in this area so correct me if needed) using a
published API so that any web-based plugin could be written to commutate
to check_by_http using this standard API.

The real problem lies in code duplication. The current check_httl should
be made into functions that are callable by check_by_http too.
Alternatively, it had been suggested we could use curl (and likely move
check_http to use curl as well), which would let us forget about the
details of the http protocol and concentrate on coding functionality
instead.

> This plugin only will work when http is functioning correctly on the 
> server the file is hosted on, and will assume this is always the case. 
> The server you are connecting to should be monitored by check_http.

Obviously - connection problems should default to UNKNOWN, overridable
by a switch (some people want to know if the http-based service fails
too). I'd like to have this default behaviour for check_nrpe, check_nt,
check_snmp and check_by_ssh too (it's likely the case already for some
of these plugins), in a major release.

- --
Thomas
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFI2FC76dZ+Kt5BchYRAsF0AKDjIyEHhkId76KWgxwsZlwMMEPjMQCgoxXD
LHY61I8J4lxKYjeODiCiFbQ=
=Rh2K
-----END PGP SIGNATURE-----




More information about the Devel mailing list