[ASK] What does dpinger result trigger please ?

Sopalajo de Arrierez

This is my test, heper:

luis@Balanceador:~/Temporal/DPinger$ cat DPinger.pfshell
pfSense shell: print_r(get_dpinger_status(WAN_DHCP));
pfSense shell: exec

luis@Balanceador:~/Temporal/DPinger$ pfSsh.php playback DPinger.pfshell

Starting the pfSense developer shell....

Fatal error: Call to undefined function readline_completion_function() in /usr/local/sbin/pfSsh.php on line 162
PHP ERROR: Type: 1, File: /usr/local/sbin/pfSsh.php, Line: 162, Message: Call to undefined function readline_completion_function()luis@Balanceador:~/Temporal/DPinger$

Am I missing something?
My shell is Bash, I don't know if it bother.

deajan

Hello,

@heper, thanks for the tip, but actually the pfSense shell won't give you more info than you could directly read from the dpinger socket.
In the following example I don't see if P1WAN_DHCP is part of my LB1 gateway group or not.


pfSense shell: print_r(get_dpinger_status(P1WAN_DHCP));
pfSense shell: exec
Array
(
    [loss] => 0
    [latency_stddev] => 0.282
    [latency_avg] => 5.918
    [gwname] => P1WAN_DHCP
    [srcip] => XX.XX.XX.XX
    [targetip] => YY.YY.YY.YY
    [status] => none
)

I still need to find the code that triggers the "add to gateway group" and "remove from gateway group" behavior.
I found the code but I'm not sure that it's the best entry point to add my code as it's a bit of a hack:
https://forum.pfsense.org/index.php?topic=109000.msg628747#msg628747

I actually search for a function that's triggered where I could add my arbitrary code.

luckman212

deajan did you find your answer yet?

I am struggling to solve the same problem as you. I understand dpinger is supposed to trigger /etc/rc.gateway_alarm on both DOWN and UP scenarios, but in my testing I am finding that it rarely (if ever) triggers during the WANUP (in which case arg $3 aka "alarm_flag" should == "0"

I added the following line to rc.gateway_alarm to add some additional logging, and then simulated a failure by yanking the Ethernet cables from my WAN1

/usr/bin/logger -t gwstate "rc.gateway_alarm event detected [$GW] mon_IP:$2 alarm_flag:$3"

I then use clog -f to track the events:

clog -f /var/log/system | grep gwstate

But while it does show alarm_flag:1 events, the corresponding alarm_flag:0 are not being triggered.

??? :o

dennypage

I expect you will find that there is a new instance of dpinger running. A restart of dpinger is probably being triggered externally before dpinger declares an alarm state change.

luckman212

@dennypage:

I expect you will find that there is a new instance of dpinger running.

Thanks Denny. How would I tell that? compare PIDs before & after? So that is not how it is supposed to work, right? I mean, the dpinger processes should not be killed and respawned just due to a hotplug event, right?

dennypage

Yes, you can check the PID to see if a new dpinger instance has been started. You will also find a startup message in the system log each time dpinger is started. Dpinger does not log on exit, so it would not be unusual to see a sequence of dpinger start, dpinger alarm on, dpinger start in the system log.

pfSense restarts interfaces and/or packages in many circumstances that at first are surprising. Most make sense after you dig into them a bit.

deajan

Hi guys,

Sorry bieng late for the party, went into holidays.
As for my answer, I modified /etc/inc/gwlb.inc in order to reach my goal.
I've created a PR on github, but Renalto said there should be a better way.
See the PR discussion here: https://github.com/pfsense/pfsense/pull/3093

luckman212

I know the dpinger events are trapped by rc.gateway_alarm but what about physical link state changes (i.e. cable is plugged/unplugged)? Is there a place we can hook into to run code after a linkup event?

deajan

As far as my testing got, rc.gateway_alarm does only trigger when a link goes down, but not up.
Physical link state changes are indeed detected as dpinger won't be able to ping whatever you configured.

My #PR solution detects link up / downs.

luckman212

Hmm ok yes I was able to log dpinger triggering rc.gateway_alarm during WAN UP events as well but it wasn't consistent. I believe as Denny said, sometimes other processes or scripts are killing dpinger and restarting it and thus it doesn't trigger the call to rc.gateway_alarm. I haven't had a chance to review the code in your PR but I will look at it.

I know Renato wants to do things "right" - of course that is always best but sometimes when the SHTF you gotta do what you gotta do.