Apinger giving optimistic RTTs in 2.2.

DemonfangArun

amusingly mine's the opposite of the problem in that thread, my pings should be 20-40ms and drop to crap (as much as i wish my ping was 1ms, that's impossible on adsl).

if they don't plan to fix or change anything perhaps someone should develop a replacement for apinger, given what it does it shouldn't take more than a week or two to write something new and get it stuffed into pfsense as an additional package.

justincase

I had the same issue, that is, every time a MONITOR would reach high latency, crap RTT.

In my case, this is a new behavior in 2.2 and it is definitely a NO GO as I badly need load balancing with apinger.

So, I ended up restarting apinger when monitoring alarm goes off:

1. a PHP script that relaunch apinger, located in "/root/restart_apinger.php":


#!/usr/local/bin/php -f
require_once("service-utils.inc");
service_control_restart("apinger", "");
?>

Don't forget to have PHP script executable:


chmod +x /root/restart_apinger.php

2. a patch in "/etc/inc/gwlb.inc" where "/var/etc/apinger.conf" is stored.

Always backup first:


cp -p /etc/inc/gwlb.inc /etc/inc/gwlb.inc~

Patch "/etc/inc/gwlb.inc" alarm command off so it run "restart_apinger.php" script on MONITOR alarm off event + some time:


alarm default {
        command on "/usr/local/sbin/pfSctl -c 'service reload dyndns %T' -c 'service reload ipsecdns' -c 'service reload openvpn %T' -c 'filter reload' "
        command off "/usr/local/sbin/pfSctl -c 'service reload dyndns %T' -c 'service reload ipsecdns' -c 'service reload openvpn %T' -c 'filter reload' ; sleep 10 && /root/restart_
apinger.php &"
        combine 10s
}

Then run apinger restart script to update configuration file:


/root/restart_apinger.php

and verify if "apinger.conf" is updated correctly:


less /var/etc/apinger.conf

Any one with a smarter way is welcome.

luckman212

Justin that's a pretty cool workaround (thanks for posting it) but I am concerned – is apinger, such a core feature of the system, that badly broken? Maybe it would help to create a bounty to fix the underlying behavior as well as add some much needed features like monitoring multiple target IPs before declaring a gateway "down" ? I for one would happily donate towards such an effort.

Harvy66

The problem is recreating the issues for the people who are able to fix it.

justincase

Pretty sure it's not that hard to re-create the issue (high latency / host down) using ipfw for example:
http://info.iet.unipi.it/~luigi/dummynet/

Not sure apinger is that badly broken: while it is/was a show stopper to me in version 2.2 it used to work just fine before. Don't blame the dev so hard, once they/we find the real problem cause, it should probably be easy to fix.

EDIT:
Humm OK ~~https://github.com/Jajcus/apinger/blob/master/BUGS~~

luckman212

Yes it seems unless I am reading it wrong that the last actual code change commit (76b1470) was 9 years ago. Maybe that needs some dusting off? :o

doktornotor

This is not what's used on pfSense.

luckman212

Oh I see- where is the source code for the apinger.c that is used in pfsense?

doktornotor

https://forum.pfsense.org/index.php?topic=76132.0

luckman212

Ok was not aware. Reading up now, thank you

justincase

So the fix I posted earlier works, but apinger still crash sometimes, so I wrote a watchdog script to put in the crontab.

The "watch_apinger.sh" shell script will look for apinger PID using ps CLI, look for apinger PID using its PID file (in case of SIG_KILL), compare both values, and run "restart_apinger.php" script if needed:


#!/bin/sh

RELAUNCH=0
CUR=$(ps xcopid,command | awk '/apinger/ {print $1}')
PID=$(cat /var/run/apinger.pid)

[ -z "$CUR" ] && RELAUNCH=1
[ -z "$PID" ] && RELAUNCH=1
[ "$CUR" != "$PID" ] && RELAUNCH=1

[ "$RELAUNCH" == 1 ] && ( killall apinger ; /root/restart_apinger.php )

Save it to "/root/watch_apinger.sh" and make it executable :


chmod +x /root/watch_apinger.sh

Then add it to root crontab:


echo "* * * * * /root/watch_apinger.sh" | crontab -