Modified Ping routine compensating for micro seconds [Split]

[mszarek]

Hi,

First of all I want to admin that I didn't check all archives to find answer regarding my question. Forgive me that - I just don't have enough time for playing with Cacti and forums.

I have found post http://forums.cacti.net/about8168.html explaining the meaning of letters next to values on graphs.
But tell me please if one can get rid of this letters and force Cacti to display more "reader friendly" values in form of: 0.234 (instead of 234 m).
I would like to get such markings for example in my ping latency graphs.

Thank you in advance for any kind of help!

Greetings

[N3NCY]

I don't know how to make is get rid of the letters either.
PINGs under one (1) msec never displayed correctly for me.
I made my ping script always round up to 1 if less than 1ms
Example:
.407us becomes 1ms
I don't care about microseconds, milliseconds is close enough for me!

Here is what my UNIX - Ping Host: cacti/scripts/ping.pl looks like now:

Code: Select all

#!/usr/bin/perl 

open(PROCESS, "ping -c 1 $ARGV[0] | grep icmp_seq |");
$ping = <PROCESS>;
close(PROCESS);
$ping =~ m/(.*time=)(.*) (ms|usec)/; 

$output = $2;
chop ($output);
chop ($output);
chop ($output);
chop ($output);

if ( $output =~ 0 ) { $output = $output + 1 };

print $output;

[tgk]

To get rid of the letters (milli, Kilo, etc), just create a new GPRINT (Graph Management/GPRINT Presets), that looks like this:

Code: Select all

%8.2lf

(without the final %s that the "Normal" preset has)

... and then just apply that preset to your labels in the Graph Template.

[Yesser]

Hi:
I have a doubt about the meaning of some results.
How can I interpret the result of a chart about errors and discards in a router's interface?
Is something like this:

Errors in Current: 0.00 Averange: 545.88 u Maximum: 94.0 m
Discards out Current: 9.33 m Averange: 2.30 m Maximum: 236.0 m

I hope you can help me...

[sauron]

Using a GPRINT preset to remove the letters works fine for the legend - but is there any way of disabling it for the Y-axis value? I've got a load average graph where I'd like the axis display to be prevented from turning into 500m - 750m, and displayed as 0.5 to 0.75.

[new-to-linux]

Thanks N3NCY

I added you code which work great for devices under 1 millisecond but for the normal Wan devices they stopped graphing. Any ideas.

Once again thanks for your help

[N3NCY]

Try running the script from the command line like:
./ping.pl devicename
or
./ping.pl WANipAddress

Show me the output.

You may need to make your script executable via
chmod +x ping.pl

[new-to-linux]

Hi N3NCY

I tried this with no luck. I made ping.pl executable via command line and ran it but it does not show me the results it just goes back to the shell line.

Sorry I could not work out how to paste a picture into post.

[rony]

This is not a discussion to have on a sticky thread.

[pdestefanis]

That script won't work if you need to graph host that are about 1ms and below 1ms.
The problem lays in that you remove the last four characters, and in some cases ping will return something like "3.78" (meaning miliseconds) and thus you'll end up with an empty return string.

You can change it to:

#!/usr/bin/perl

open(PROCESS, "ping -c 1 $ARGV[0] | grep icmp_seq |");
$ping = <PROCESS>;
close(PROCESS);
$ping =~ m/(.*time=)(.*) (ms|usec)/;

$output = $2;

if ( $output =~ 0 ) { $output = $output + 1 };

print $output;

Wich is not a nice hack, since all it does is adding 1ms when the RTT is < 1ms. Horrible as it is, it will work with both.

A nicer approach will be to check if the first digit is 0 (as in 0.234 ms) and the how many digits you have after your decimal, and get rid of those and add 1... I don't have the time, though)