Multiwan problems

aseques

Good morning,

I am on pfsense 1.2.2, and I've tested nearly everything I can, I hope there's someone that can shed some light to this..
This is the scenario:

Two dsl routers with nat disabled (pfsense has both public IPs)
A Wan interface for the users.

The Problem:

Mar 10 10:30:14 	check_reload_status: reloading filter
Mar 10 10:30:12 	slbd[7186]: Service Balance changed status, reloading filter policy
Mar 10 10:30:12 	slbd[7186]: Service ADSL0backADSL1 changed status, reloading filter policy
Mar 10 10:30:12 	slbd[7186]: ICMP poll succeeded for 123.456.789.123, marking service UP
Mar 10 10:30:12 	slbd[7186]: ICMP poll succeeded for 123.123.123.123, marking service UP
Mar 10 10:30:08 	check_reload_status: reloading filter
Mar 10 10:30:07 	slbd[7186]: Service Balance changed status, reloading filter policy
Mar 10 10:30:07 	slbd[7186]: Service ADSL0backADSL1 changed status, reloading filter policy
Mar 10 10:30:07 	slbd[7186]: ICMP poll failed for 123.456.789.123, marking service DOWN
Mar 10 10:30:07 	slbd[7186]: ICMP poll failed for 123.123.123.123, marking service DOWN

Sometimes, not due to traffic (now the pfsense is back to my testbed), pfsense thinks that both ips are down at the same time wich is not true according to the router.
After this event the connection is down for 20 second causing problems to everyone, specially me :(

I've tested so far:

D-Link DFE-580TX (freeBSD supported) card as a NIC
Two elder 3COM 3C905 Cards instead of the D-Link
Pools with balancing / only failover
Tested various upstreams, including using the same-net router ip
Modified vslb.inc to 30000 not to send to much ping to my upstreams

tkadams

Are both DSL circuits from the same provider?

I had a similar issue with 2 cable modems. It seemed that there was arp table instability issue since the provider's server's MAC address showed up on both of the WAN interfaces. My solution was to install another NAT device in front of one interface.

Could you let one of your modems hold the pppoe info as well as NAT?

Hope this helps…

Tim

aseques

Could you let one of your modems hold the pppoe info as well as NAT?

In fact the two DSL are taking care of the PPPoE stuff, the only change I did to from the default router setup is to disable NAT and create static routes, so I don't have to cope with the limit of my little zyxel's nat tables.
I'll now have a look to the arp routes and see also if I can do the PPPoE from the pfsense too ..

Thanks for your help

xelor

Same Problem here,

Only difference in my config - I use

one WAN with PPPoE (with NAT) - adsl2+

one WAN with static IP (LLC encapsulation) - router is only bridging connectivity providing ATM syncing with ISP allowing pfsense act as edge device

Running on the multicore enviroment ( 7.0-RELEASE-p8 FreeBSD 7.0-RELEASE-p8 #0: Thu Jan 8 22:29:30 EST 2009 sullrich@freebsd7-releng_1_2_1.pfsense.org:/usr/obj.pfSense/usr/src/sys/pfSense_SMP.7 i386)

Interfaces in approx 20min going down for 30 seconds marking one of the WANs interfaces down.

aseques

@xelor:

Same Problem here,

one WAN with PPPoE (with NAT) - adsl2+

one WAN with static IP (LLC encapsulation) - router is only bridging connectivity providing ATM syncing with ISP allowing pfsense act as edge device

Oh interesting, finally someone with the same problem!

A couple of tips that might help you:
.- PPPoE and sticky connections are not friends before pfsense 2.0, you shouln't use them
.- To get the version the best place is to go to Status -> System , in there you can see the version with build time.
.- Update to 1.2.3, this version seems to solve a lot of issues, not mine but even tough.

Finally, I would like to know what hardware are you using (board + network interfaces)

Cheers

xelor

Oh interesting, finally someone with the same problem!

A couple of tips that might help you:
.- PPPoE and sticky connections are not friends before pfsense 2.0, you shouln't use them
.- To get the version the best place is to go to Status -> System , in there you can see the version with build time.
.- Update to 1.2.3, this version seems to solve a lot of issues, not mine but even tough.

Finally, I would like to know what hardware are you using (board + network interfaces)

Cheers

Hi,

Sorry for the delay in answer I was extremely busy in last few weeks and completely forgot about this post.
Anyway I have made some more testing regarding this issue and here what I managed to learn:

Switching WAN from PPPoE to Static helps a little bit - interface are not dropped so often - however ramdom downtimes for 10 sec. are still occuring approx. once every 2 days
I did not test it with 1.2.3 version - and to be honest I do not really have a time at the moment. Changing router config was quicker :-D
Regarding HW - here is my hardware profile:
Dell PowerEdge sc 440 (dual-core pentium x86_64)
motherboard - dell branded for this server model
NICs - 5 x BROADCOM 5721 GIGABIT ETHERNET ADAPTER