How restart OpenVPN server

[sullrich]

At this point I am at a loss.  Will have to discuss it with the other devs.  We are all really confused on this one.

[tpepels]

Same problem on my box ???.

root    lighttpd  1785  10 tcp4  *:1194                :
root    check_relo 339  10 tcp4  *:1194                :

Hope you find the problem soon, good luck anyway!

[Bredys]

Still nothing new about this problem ? I try every snapshot but without any progress :(

[Selective]

The only thing you can do is to make your changes and save, click the disable box to disable tunnel and then restart pf, and when its up again, click box to enable tunnel again.

[trendchiller]

Same problem here and at a friends system and at work, too… even switching to another port did not work (only for one day - using 1195 now) and the system at work... still no changes  :'(

[thinair]

I have the same issue (and have had for a while now), the OpenVPN server tells me whatever port number I'm using is already in use.  I've tried with the latest snapshot (Jan 7/06), same issue.

[sullrich]

Known issue. It's covered in 3-4 other threads but there is no solution as of yet.

[wdbacker]

I'm having the same problem with my server in UDP mode. TCP mode works perfectly for me. Looking at the listening server processes with "sockstat -l" reveals:

_dhcp    dhclient  794  10 udp4  *:1194                :
root    dhclient  697  10 udp4  *:1194                :

Apparently, the dhclient process is listening on UDP port 1194 …  ???

FYI, my box is connected at the WAN side through DHCP to my ISP. In the OpenVPN server, I enabled dynamic dns clients.

[sullrich]

There is some kind of bug where processes are inheriting other socket descriptors.

[wdbacker]

Thanks for the information Scott!

I did some more testing and I saw the same problem now with the OpenVPN server in TCP mode. Hence I think it doesn't matter if the connection is through TCP or UDP, the same problem shows up. Rebooting solves the problem. The problem also seems to happen at random.

If there is anything I can do to help you finding the problem (socket descriptors being reused?), I'll be happy to do more testing!

[wdbacker]

Hi Scott,

I noticed your Check-In 16202 on the CVS trac and I modified my /etc/inc/filter.inc as shown. Now in my case, OpenVPN is again (re)starting normally without the socket descriptors being reused by other processes! It works in both TCP and UDP server mode now (I use TCP for roadwarriors and UDP for site to site).

I will do some more extensive testing one of these days.

Thanks for the nice solution! :)

[sullrich]

Great!  Glad to hear that it has solved the issues.

[thinair]

I'm testing the updated filter.inc file as well.  I'll let you know in 24h if the OpenVPN tunnel is still up.  It usually dies after a couple hours for me.

[Bredys]

I tried last snapshot from 22.01.07 and openVPN work great !

Thanks for this fix !

[Selective]

it´s working for me aswell !!! :D

[trendchiller]

Y E A H

Scott you rule !

Happy to see this bug to be gone  ;D

[thinair]

Yep, everything is A-OK for me too.  The tunnel didn't go down once.  I guess I'll just install the 1-22-07 snapshot now.

[sh_man]

I need to implement this solution for a couple of temporary sites I am setting up for a weeks time :(.

I need to run them from a LiveCD.

How can I create a LiveCD with this fix in.

Please note that I am predominately a windows man, don't have much BSD/Linux experience and no machines other than a production firewall running pfSense :).

[hoba]

Just fetch a new livecd from the snapshotserver instead a 1.0.1 release from the mirrors that has this bug fixed: http://snapshots.pfsense.org/FreeBSD6/RELENG_1/

[sh_man]

Thanks - I'll see how I get on with that.