Update pfsense 2.0.1 stable to 2.1 problem with routes

[maykel535]

Is more,

I shutdown pfsense 2.1, and I turned on pfsense 2.0.1 I had a backup before performing the upgrade. And look …

root@pre:~# ping 192.168.1.20
PING 192.168.1.20 (192.168.1.20) 56(84) bytes of data.
From 10.100.100.3: icmp_seq=1 Redirect Host(New nexthop: 10.100.100.4)
64 bytes from 192.168.1.20: icmp_req=1 ttl=127 time=28.7 ms
From 10.100.100.3: icmp_seq=2 Redirect Host(New nexthop: 10.100.100.4)
64 bytes from 192.168.1.20: icmp_req=2 ttl=127 time=28.3 ms
From 10.100.100.3: icmp_seq=3 Redirect Host(New nexthop: 10.100.100.4)
64 bytes from 192.168.1.20: icmp_req=3 ttl=127 time=29.8 ms
^C
--- 192.168.1.20 ping statistics ---
3 packets transmitted, 3 received, 0% packet loss, time 2002ms
rtt min/avg/max/mdev = 28.399/28.989/29.840/0.616 ms
root@pre:~# traceroute 192.168.1.20
traceroute to 192.168.1.20 (192.168.1.20), 30 hops max, 60 byte packets
1  10.100.100.3 (10.100.100.3)  0.577 ms  0.600 ms  0.631 ms
2  10.100.100.4 (10.100.100.4)  0.766 ms  0.826 ms  0.897 ms
3  192.168.1.20 (192.168.1.20)  31.176 ms  31.609 ms  31.779 ms

came perfectly to 192.168.1.20

BUG PFSENSE 2.1 ROUTES???

[maykel535]

http://imageshack.us/photo/my-images/30/ynvc.png/

[ptt]

@maykel535

Not related to your issue, but is better if you attach the screenshots directly to your post reply, instead of imageshack ;)

[maykel535]

Worth, and you have to see what these commenting me photo to the problem that I have … There is a piece of bug in pfsense, I assure you.

[maykel535]

For the moment and to take my headaches, I have set up the backup pfsense routes 2.0.3 stable and work well. Until we solve this BUG, we will continue with this version.

[stephenw10]

@maykel535:

root@pre:~# traceroute 192.168.1.20
traceroute to 192.168.1.20 (192.168.1.20), 30 hops max, 60 byte packets
1  isp (ip public isp)  1.534 ms  1.592 ms  1.611 ms
2  isp (ip public isp)  2.107 ms  2.199 ms  2.234 ms
3  192.168.66.121 (192.168.66.121)  2.747 ms  2.847 ms  2.868 ms^C

Where is the machine in this traceroute, 192.168.66.121? That is also in the 192.168.0.0/16 subnet. It's not surprising pfSense is getting confused if it can access that subnet via both gateways.

Why does ping work but traceroute doesn't.  :-\ Edit: Misread that.

Steve

[maykel535]

Ip that is not ours, I teach a traceroute goes from pfsense 2.0.3:

[2.0.3-RELEASE] [admin@pfsense-mo2o-ketchum.mo2o.com] / root (1): traceroute 192.168.1.20
traceroute to 192.168.1.20 (192.168.1.20), 64 hops max, 52 byte packets
  1 10.100.100.4 (10.100.100.4) 1,020 ms 0.615 ms 0.586 ms
  2 * 192.168.1.20 (192.168.1.20) 28.836 ms 37.787 ms

All ok not see? While the traffic routes. Pfsense 2.1 The problem is that when I try to get from an internal host to the ip 192.168.1.20 instead of route to 10.100.100.4, I routed to our wan. What makes it completely wrong.

It's the same rule for pfsense 2.0.3 to 2.1 that is more I tried to delete it and recreate it. Hallucinate with pfsense 2.1 not understand anything. I assure you the route is straight.

Probe a facility from 0 to pfsense 2.1 stable instead of doing an upgrade, and import the configuration xml, yet not strip.

I do not know what else to do, so as a need for programmers routes, will continue with pfsense 2.0.1 until they fix the bug, although they say it's not a bug.

Sorry for my english.

Thanks for all, true.

[stephenw10]

It seems very strange to me that you have a private subnet somewhere upstream of your public WAN IP.

You need to look at the pfSense routing table and see if it has somehow acquired a route to 192.168.. on the WAN. Perhaps via some new routing protocol introduced in 2.1.

Normally if you had specified a gateway in the firewall rule, and that rule is actually catching the traffic, then it can only use that gateway. However there is a background rule that will by-pass that for local networks the 'negate rule'. You can turn it off in System: Advanced: Firewall/NAT. That also existed in 2.0.3 but perhaps it didn't recognise it as local there.

In 2.0.3 do you need to have a firewall rule specifying a gateway?

Specifying a gateway on the same interface that the traffic arrives is the sort of thing that can fail to work. See NAT reflection for example.

Steve

[maykel535]

I spent alone with the IP 192.168.1.20, it happens with all routes. A little weird right? Indeed, for fanjar this topic I say I've tried to add the same routes to pfsense 2.1 BETA and I have worked flawlessly. Pfsense 2.1 if stable has a bug or if, or add something new to fuck routes because it does not work any of the 4 that I have on pfsense 2.1 stable. There is a bit weird that I work perfectly in pfsense routes 2.0.1 stable and 2.1 BETA!

Thanks for everything.

[tpramos]

Same problem her. In 2.0.3 routes work fine, in 2.1 do not!
Test upgrading in place and fresh install, both not work.

[stephenw10]

The same sort of static routes to an internal gateway?

Steve

[tpramos]

Yep! two Pfsense with Wan in same network.

[stephenw10]

Can you specify the routes and under what conditions they don't work? For example maykel535 was able to use the route from the pfSense box itself but not from a client in the same subnet as the gateway.

Steve

[maykel535]

do not compliqueis more, I can not lose more time with this because in pfsense 2.1 BETA works. Something as simple as a route, I can not lose more time. There is a bug and I just want to help you solve … A shame since pfsense 2.1 stable comes with a amount of improvements as MultiWAN dynamic dns and a bunch of bug fixes.

The real problem is that instead routes of enrutarte to the host that is, it takes you out to the wan and not because there is anything wrong, but simply routed bad this version of pfsense 2.1

We will have to wait ...

[__Fox__]

any officiale update or replay to this problem?
Upgrading to 2.1 in multi WAN-LAN configuration is a hell!

[tim.mcmanus]

I too had some routing issues from 2.03 -> 2.1.  I think 2.0.3 was more forgiving with routes than 2.1.  And interestingly enough, I too am running a multi-WAN.

In my configuration I have two LANs and two WANs.  I want each LAN to go out a different WAN, so I created a rule with each gateway in it.

When this is set up in 2.0.3, LAN to LAN2 communication will occur without a specific rule to route traffic between them.  However, I had to specifically create a rule in each LAN to route traffic between them or else it would not happen.  I assume this is due to creating a rule specifying which gateway each LAN needs to use.  It was pretty frustrating for a few week since the LAN serves resources up to LAN2 and vice versa.

I've attached screen shots of my new rules.  But as I mentioned before, an upgrade from 2.0.3, which was working flawlessly, to 2.1 required me to troubleshoot and create new explicit rules to route traffic between my two LANs.

So for folks experiencing issues, I would recommend that you sit down and think about rewriting your rules and validating them.  I had to do hours of testing to ensure that my new rules worked and didn't break anything else.


[gilmarcabral]

Using OpenVPN tunnel with ovpns1, and created a route does not accept ipv6 automatically switch to ipv4 and remove.
Was not done the update version 2.0.3 is rather a new installation

[__Fox__]

Hi Tim,
  I just added the crossing rules from LAN1 to LAN2 and viceversa, and I put them before the rule that send the traffic to a specific routing group (otherwise the intranet traffic goes out to the gateway!?!?!) but I still have problem… I will try to reset the configuration and restart, but I  have 40 rules and nat with 6 interfaces, I never had problems in upgrading before  :-(

Alessandro

[tim.mcmanus]

The issue I was experiencing was limited to the two LANs communicating.  I have many rules on my WANs, but my LANs are rather simple (by design).  Have you done any troubleshooting to know specifically where the issue is occurring and which routes are affected?

I always start with a ping and traceroute, and then check the firewall logs.  I can usually determine rather quickly why things aren't getting from point A to point B using those three tools.  If you need assistance interpreting those things, just post the data here.  Another suggestion; I usually time my testing on 5-minute intervals so I can match up different logs with one another.  Sometimes if you test, make a change, and test again, you can miss things in the logs.

[stephenw10]

The behavior of policy based routing, which is what we're dealing with here, changed between 1.2.X and 2.0.X. At that time automatic 'negate' rules were added behind the scenes that allowed traffic to negate a policy based rule when the destination is a local subnet. I was burnt by that change because I had rule sets that had previously not allowed traffic between local subnets that now suddenly did. Worse, I didn't realise anything had changed for some months. That'll teach me to read the release notes properly.  ::) Anyway after some requests a check box to disable the negate rules was added in System: Advanced: Networking: I always have it unchecked, I don't want the negate rules catching me out again. I wonder if the behaviour changed again in 2.1, it seems to fit the symptoms.

Steve