[smokeping-users] CiscoRTTMonEchoICMP probe and IOS 15.x

[Tobias Oetiker]

Hi Mateusz,

iiks, this is a bit twisted I see ... without going into brain
surgery, how about this:

Index: CiscoRTTMonEchoICMP.pm
===================================================================
--- CiscoRTTMonEchoICMP.pm      (revision 1087)
+++ CiscoRTTMonEchoICMP.pm      (working copy)
@@ -59,7 +59,7 @@
 The above configuration grants SNMP read-write only to 10.37.3.5 (the smokeping host) and only to the ciscoRttMon MIB tree. The probe does not need access to SNMP variables outside the RttMon tree.
 DOC
                bugs => <<DOC,
-The probe sends unnecessary pings, i.e. more than configured in the "pings" variable, because the RTTMon MIB only allows to set a total time for all pings in one measurement run (one "life"). Currently the probe sets the life duration to "pings"*2+3 seconds (2 secs is the ping timeout value hardcoded into this probe).
+The probe sends unnecessary pings, i.e. more than configured in the "pings" variable, because the RTTMon MIB only allows to set a total time for all pings in one measurement run (one "life"). Currently the probe sets the life duration to "pings"*5+3 seconds (5 secs is the ping timeout value hardcoded into this probe).
 DOC
                see_also => <<DOC,
 L<http://oss.oetiker.ch/smokeping/>
@@ -75,7 +75,7 @@
        }
 }

-my $pingtimeout =2;
+my $pingtimeout = 5;

 sub new($$$)
 {
@@ -274,6 +274,12 @@
 execute the pings, as well as the SNMP community string on the router.
 DOC
                },
+        timeout => {
+             _re => '\d+',
+             _example => 15,
+             _default => $pingtimeout+10,
+             _doc => "How long a single RTTMonEcho ICMP 'ping' take at maximum plus 10 seconds to spare. Since we control our own timeout the only purpose of this is to not have us killed by the ping method from basefork.",
+        },
                iosint => {
                        _example => '10.33.22.11',
                        _doc => <<DOC,


the idea is to set the timeout 'per ping' to pingtimeout+5 seconds
... this may be a bit of a long timeout but since we control our own time anyway it will never hit.

cheers
tobi

Yesterday Mateusz Blaszczyk wrote:

> Tobi,
>
>
> >>
> > [...]
> >
> > I don't have such a router at my disposal, but I'll be glad to
> > integrate your patch ... not sure what you mean with the timeout of
> > 100s ... this seems a rather long time to me ... especially as you
> > mention $pingtimeout=5.
> >
>
> well - I am talking about this (probe timeout, not target timeout)
>
> probes/basefork.pm:168
>         # we add 1 so that the probes doing their own timeout handling
>         # have time to do it even in the worst case
>         my $timeout = $maxpings * $maxtimeout + 1;
>
> 20 pings * 5s  + 1 = 101 so default $timeout==5 for probe is not enough
>
> So that's why I have suggested $timeout for probe == 10.
>
> there are a lot timeouts here :)
>
> > how would your complete patch look like ?
> >
>
> the probes/basefork.pm is a generic file for all probes so I assume
> that only the one that I have posted before. Not sure if I can specify
> the default probe timeout in each probe source file?
>
>
> --- /tmp/CiscoRTTMonEchoICMP.pm       2010-03-08 10:17:18.000000000 +0000
> +++ CiscoRTTMonEchoICMP.pm    2010-03-05 15:53:39.000000000 +0000
> @@ -75,7 +75,7 @@
>       }
>  }
>
> -my $pingtimeout =2;
> +my $pingtimeout =5;
>
> sub new($$$)
> {
>
>
> Best Regards,
>
> -mat
>
>

-- 
Tobi Oetiker, OETIKER+PARTNER AG, Aarweg 15 CH-4600 Olten, Switzerland
http://it.oetiker.ch tobi at oetiker.ch ++41 62 775 9902 / sb: -9900