Forum Discussion

sosa123_86561's avatar
sosa123_86561
Icon for Nimbostratus rankNimbostratus
Jun 22, 2008

F5 and RTCP

hi i am new in F5 , we need some solution to fix this issue

our date: 
 
    
  CLIENT: ---------  
                   |  
                   |----  
                        |  
  VS01:1.1.1.18:554     LTM------------------ (Ext) External RTSP server 20.20.20.220:554  
  Pool:abc_554          |SNAT(1.1.1.18 --192.168.1.25 and 192.168.1.26)  
  VS02:1.1.1.18:0       |  
  Pool:abc_udp          |  
                        |  
              ---------------------  
             |                    |  
           NODEA                NODEB  
          192.168.1.25:8443     192.168.1.26:8443  
          GW: LTM               GW:LTM

Client connect to VS01 then rtsp activity start with for example NODEA.

NODEA send this request to (Ext) server

Ext Server response back to NODEA and then

NODEA Respond back to Client

so fare every think is as expected

when RTP or RTCP start from CLIENT site after successful completion of RTCP,the request goes to NODEB instead on NODEA( where the connection was first initiated for RTSP ) of course destination un reachable happened , and then it when request come back to pool member NODEA as i send before RTP or RTCP working fine .

Our issue is how-to sync both VS in such a way that every RTCP or RTP request goes to its RTSP connection NODE.

We have persistence profile for VS01` in place.

Hopefully some one will answer

config
design

5 Replies

Sort By
  • Hamish's avatar
    Hamish
    Icon for Cirrocumulus rankCirrocumulus
    Jun 23, 2008
    Will match across services or match across virtual servers on the persistence profiles for VS01 & VS02 do what you require?

     

     

  • Deb_Allen_18's avatar
    Deb_Allen_18
    Historic F5 Account
    Jun 23, 2008
    Will match across services or match across virtual servers on the persistence profiles for VS01 & VS02 do what you require?

     

     

    It may, but it wasn't clear from the info sosa provided how the pools are configured for each VS.

     

     

    You can only persist across virtuals/services if the pool members are the same in both cases (IP+port must match). Otherwise the system cannot find the "same" pool member in each pool.

     

     

    /deb

     

     

  • sosa123_86561's avatar
    sosa123_86561
    Icon for Nimbostratus rankNimbostratus
    Jun 23, 2008
    pool configuration is as under:

     

     

    VS01:1.1.1.18:554

     

    Persist profile apply only VS01

     

    Pool:abc_554

     

    loadbalance mathod = roud robin

     

    Node01 :192.168.1.20 port 8443

     

    Node02 :192.168.1.25 port 8443

     

     

    ===========================

     

     

    VS02:1.1.1.18:any

     

    Pool:abc_udp

     

    loadbalance mathod = roud robin

     

    Node01 :192.168.1.20 port any

     

    Node02 :192.168.1.25 port any

     

     

    ======================

     

     

    fisrt of all thanks for give me reply , yes the pool members are same only the port on both pool are different. waiting for your reply.

     

     

     

    BR

     

    sosa

     

     

  • Deb_Allen_18's avatar
    Deb_Allen_18
    Historic F5 Account
    Jun 25, 2008
    A pool member is defined as IP+port, so the pool members are actually NOT the same between pools.

     

     

    The solution here would be to use the port 8443 pool for both virtual servers, with port translation enabled on VS01. (VS02 will not translate the port regardless of whether translation is enabled since there is no port to translate to.)

     

     

    The big question then becomes how to service check the pool members for not only the port 8443 traffic, but the services that are behind VS02. The answer is easy: Create multiple monitors that test all required services, and apply them all to the pool members, requiring all to succeed before pool member is UP.

     

     

    HTH

     

    /deb
  • Deb_Allen_18's avatar
    Deb_Allen_18
    Historic F5 Account
    Jun 26, 2008
    Colin & I were talking this over for our Post of the Week recording, and I realized I had recommended using the wrong single pool -- I have corrected my previous post.

     

     

    Also wanted to re-iterate that the persistence profile with "Match Across Virtuals" must be applied to both virtual servers ("Match Across Services" might work in your case instead since the VS address is the same)

     

     

    Will post back with the Post of the Week link when it gets published next week. In the meantime, post back if you have questions.

     

     

    /deb