TCP & HTTP Server Response Time: Script & Templates

[kayalinux]

Here is a another contribution.

I implemented TCPResponseTime and HTTPResponseTime to measure the response times of aTCP port and of HTTP web page. The Perl script launches queries repeatedly.

Upon import, the attached .XML file will create both a Data Template and a Graph Template called "TCP Response Time" "HTTP Response Time" for Data Source, as well as a Data Input Method with the same name.

The Graph Template as no special features see attached screenshot.

In order to install it:
1) import xml files
2) copy perl script to the cacti's scripts directory.

[macindy]

Thank you for your scripts! Great stuff.

In my stats are often gaps - can I do something against it?

[nahun]

is the latency being reported the average latency or the current latency from the latest poll on both the TCP and HTTP scripts?

[cigamit]

[indee]

hi,

first of all thanks.

I got the graph for tcp response, bit problem is with the HTTP response, there is no output (line etc).

strange! here is my setup

cacti 0.8h with architecture plugins from (cactisusers.org) on FC 4.

thanks in advance.

Indee

[ahardman]

I get the following when I use the script.

05/09/2006 12:47:03 PM - CACTID: Poller[0] Host[11] DS[2641] SCRIPT: perl /www/htdocs/cacti/scripts/tcp_response_time.pl 64.233.167.147 80, output: U
05/09/2006 12:47:03 PM - CACTID: Poller[0] Host[11] DS[2641] WARNING: Result from SCRIPT not valid. Partial Result: ...

I seems that the result is not in the right format.

Any ideas

The script returns a result in the form avg:### when run from the command line on the server

[gandalf]

Latest cactid requires /full/path/to/perl for the Data Input Methods. Please try this and Clear Poller Cache afterwards
Reinhard

[ahardman]

The perl script was the problem Thanks

[morrowd]

How would I use this script to monitor something other than the root of the website?

ex. http://www.mydomain.com/folder/mytest.htm

[gandalf]

Take a look e.g. at the http script. You may provide 2 further parms, http_port and url (use them in the write_request statement). Then, you have to change the Data Input Method and add these two parms as input parms as well. Last, the Data Template should hold defaults for this new input parms, e.g. 80 as http_prort and "/" as url. Check the Use Per-Data Source Value (Ignore this Value) under Custom Data to be able to supply new input parms for each graph to be created.
Reinhard

[fla_roadkill]

Could someone post a modified script that show's how to use the URL and PORT? Thanks!

[parkynz]

Hi, nice template - thanks, but the latency reported includes the time taken for the client (perl script) to resolve the server's DNS name and is not really relevant when just looking at the response time of the http server itself. Can be worked around by using local hosts file (on windows) but I'm sure a perl DNS resolution of the host before capturing the start time would be smarter?

If you run a caching DNS client then no problem really but that's not always the case.

Or am I missing something here since I'm a newbie..?

cheers.
~Andy.

[guy2006]

my template with min, max,last & avg

[rwhittaker]

Question:

How will this script behave with relation to proxy servers?... Is there any way to "force" the script to use one?...

Thanks,
Richard.

[guy2006]

I made new script & tempate because the first one is not realy mesure the time till the GET request is finished.

the new script is based on nagios plugin check_http with the follow option so the time reflaced in the graph is the time taken for the page to load

copy the scripts check_http & nrpe_http to the cacti scrips directory.
Import the graph templates.