I went ahead and labbed this up but only to find what I had learned and 
committed to memory was not correct about whom initiates the BGP open session.  
:/

My ASA shows that the lower IP address sent the BGP OPEN to the higher IP.  

Jason- was your recommendation to only allow BGP from the inside to the outside 
and let the routers work it out on their own?


My Lab output:
ASA-LAB01(config)# sh conn
8 in use, 18 most used
TCP outside 200.200.200.1:179 inside 220.220.220.2:45572, idle 0:00:00, bytes 
0, flags saA

 
 
Regards,
Jay McMickle- 3x CCNP (R&S,Security,Design), CCIE #35355 (R&S)
 


________________________________
 From: Jason Madsen <[email protected]>
To: Eugene Pefti <[email protected]> 
Cc: Jay McMickle <[email protected]>; Fawad Khan <[email protected]>; 
"[email protected]" <[email protected]> 
Sent: Monday, September 3, 2012 2:43 PM
Subject: Re: [OSL | CCIE_Security] BGP through ASA
 

yep, i've always seen both BGP peers will initiate a connection to TCP 179 on 
the other, and then once a connection is established, the other one drops off.  
no need for outside ACL unless desired or specified by lab task.  In the real 
world, I consider it a best practice though.

Jason



On Mon, Sep 3, 2012 at 1:13 PM, Eugene Pefti <[email protected]> wrote:

Thanks, Jay,
>I wish it is very simple and clear. My lab routers don’t stick to the rules 
>you described.
>Let’s drop authentication from the picture and look into the mere session 
>establishment.
> 
>R5 -------(192.168.7.0)-------(inside)ASA(outside)------(192.168.6.0)------R3
> 
>On R3 I set router ID to be 200.200.200.200 to be higher than R5 ID
> 
>router bgp 103
> no synchronization
> bgp router-id 200.200.200.200
> bgp log-neighbor-changes
> network 192.168.33.33 mask 255.255.255.255
>  neighbor 192.168.7.5 remote-as 105
> neighbor 192.168.7.5 password cisco
> neighbor 192.168.7.5 ebgp-multihop 255
> no auto-summary
> 
>R5 
>router bgp 105
> no synchronization
> bgp log-neighbor-changes
> network 192.168.55.55 mask 255.255.255.255
> neighbor 192.168.6.3 remote-as 103
> neighbor 192.168.6.3 password cisco
> neighbor 192.168.6.3 ebgp-multihop 255
> 
>Then according  to you I expect R3 will initiate BGP session and it should 
>fail because I don’t have a hole in ASA for BGP traffic.
>But both peers establish the session and even though I see denies on the ASA:
> 
>ASA2# %ASA-4-106100: access-list OUTSIDE-INBOUND denied tcp 
>outside/192.168.6.3(18358) -> inside/192.168.7.5(179) hit-cnt 1 first hit 
>[0xe560841e, 0x0]
> 
>And R5 sees R1 as 192.168.6.3 not 200.200.200.200
> 
>R5#sh ip bgp sum   
> 
>Neighbor        V    AS MsgRcvd MsgSent   TblVer  InQ OutQ Up/Down  
>State/PfxRcd
>192.168.6.3     4   103   12277   10429     3703    0    0 00:12:31        3
> 
>Eugene
> 
>From:Jay McMickle [mailto:[email protected]] 
>Sent: Monday, September 03, 2012 9:02 AM
>To: Eugene Pefti; Fawad Khan
>Cc: [email protected]
>
>Subject: Re: [OSL | CCIE_Security] BGP through ASA
> 
>Gents:
>One thing to remember- the BGP peer with the highest IP (used for peering) 
>will initiate to the lower IP peer via TCP 179.  Use this to determine which 
>interface on the ASA to allow this one.  The return traffic will be stateful.  
>
>If R1 has 200.1.1.1 and R2 has 200.2.2.2, R2 would then initiate the TCP 179 
>connection.
>
>One other item when considering BGP authenticated peers through an ASA is the 
>random sequence number.  This is where most lose points on the exam.  I found 
>a quick link for reference, pasting it below.
>
>Happy to help.  Happy labbing.  ;)
>
>*Just a sample, but this is included in IPX's BLS for CCIE Security*
>http://www.packetslave.com/2009/07/12/bgp-through-an-asa-with-authentication/
>tcp-map BGP_FIX
>  tcp-options range 19 19 allow
>!
>access-list BGP permit tcp any any eq 179
>!
>class BGP
>  match access-list BGP
>  !! could also use match protocol tcp eq bgp
>!
>policy-map global_policy
>  class BGP
>    set connection advanced-options BGP_FIX
>    set connection random-sequence-number disable
> 
> 
> 
> 
>Regards,
>Jay McMickle- 3x CCNP (R&S,Security,Design), CCIE #35355 (R&S)
> 
> 
>
>________________________________
> 
>From:Eugene Pefti <[email protected]>
>To: Jay McMickle <[email protected]>; Fawad Khan <[email protected]> 
>Cc: "[email protected]" <[email protected]> 
>Sent: Sunday, September 2, 2012 9:03 PM
>Subject: RE: [OSL | CCIE_Security] BGP through ASA
> 
>I may have not be very clear or eloquent asking this question.
>Would we be punished if add a permissive BGP traffic ACL entry on the ASA 
>outside interface if the session establishes owing to the BGP peer that 
>originates it from behind the ASA?
> 
>Eugene
> 
>From:Jay McMickle [mailto:[email protected]] 
>Sent: Sunday, September 02, 2012 7:00 PM
>To: Fawad Khan
>Cc: Eugene Pefti; [email protected]
>Subject: Re: [OSL | CCIE_Security] BGP through ASA
> 
>Just remember the keyword at the end of the ACL for BGP passing through the  
>ASA. ;) (google that)
>
>Regards,
>Jay McMickle- CCIE #35355 (RS), 3x CCNP (RS,Security,Design)
>Sent from my iPhone
>
>On Sep 2, 2012, at 8:49 PM, Fawad Khan <[email protected]> wrote:
>For the exam I would do what the task say. And NOT overdo/ or over think.
>>
>>On Sunday, September 2, 2012, Eugene Pefti wrote:
>>I assume it is only for the situation when you need to control outbound 
>>traffic. For the purpose of CCIE lab should we bother with outbound ACL? It 
>>is trusted traffic per ASA security levels. 
>>
>>Sent from iPhone
>>
>>On Sep 2, 2012, at 11:13 AM, "Fawad Khan" <[email protected]> wrote:
>>The best scenario would be to have acl on both interfaces to allow 
>>communication from either side. 
>>>I would Ab inbound acl on the outside interface and inside interface.
>>>
>>>On Sunday, September 2, 2012, Eugene Pefti wrote:
>>>Hello folks,
>>>I have a rhetoric question.
>>>I believe this is a classic task when BGP peers need to authenticate through 
>>>the ASA but my question is not about it.
>>>One of my BGP peers is on outside of the ASA and the other is inside. The 
>>>ACL on ASA doesn’t allow BGP traffic from the outside peer and I see 
>>>corresponding denies when it tries to talk to the inside peer.
>>>But nothing prevents the inside peer to establish the active session with 
>>>its outside peer and they successfully do it.
>>>Now the question.  Would you add the ACL on the ASA  outside interface  to 
>>>allow BGP traffic from the outside peer to the inside one or as long as they 
>>>can establish the session that originates from the inside BGP peer we are OK?
>>> 
>>>Eugene
>>> 
>>>
>>>
>>>-- 
>>>FNK, CCIE Security#35578
>>
>>
>>-- 
>>FNK, CCIE Security#35578
>_______________________________________________
>>For more information regarding industry leading CCIE Lab training, please 
>>visit www.ipexpert.com
>>
>>Are you a CCNP or CCIE and looking for a job? Check out 
>>www.PlatinumPlacement.com
> 
>_______________________________________________
>For more information regarding industry leading CCIE Lab training, please 
>visit www.ipexpert.com
>
>Are you a CCNP or CCIE and looking for a job? Check out 
>www.PlatinumPlacement.com
>
_______________________________________________
For more information regarding industry leading CCIE Lab training, please visit 
www.ipexpert.com

Are you a CCNP or CCIE and looking for a job? Check out 
www.PlatinumPlacement.com

Reply via email to