Forum Discussion

Thrillseeker_12's avatar
Thrillseeker_12
Icon for Cirrus rankCirrus
Sep 04, 2015

Packet loss on Floating IP only???

Hi f5 community,

 

I had a very strange situation today with one of our customers. Our customer is running an BIGIP-1600 Active/Standby LTM HA setup with MAC masquerading enabled.

 

This morning the customer reported a service outage of his Exchange 2013 infrastructure running via the mentioned BIGIP-1600 LTM system (created with Exchange 2013 iAPP). After some investigation we found out that we had around 25% PING packet loss when pinging the Floating-IP, but ONLY the Floating IP, all other IP's (Self or VIP) had no PING packet losses! So it was unlikely to have NIC hardware issue on the BIG-IP systems.

 

As I mentioned already on this system we have had MAC Masquerade enabled, and this configuration is running fine for more than one year.

 

Trying to solve the issue we rebooted both HA members several times. Also we did multiple failover between the two members. But without any success...:-(

 

At the end, the only solution was to disable/enable all TMM interfaces on both HA members in parallel. After that, we haven't had any PING losses at all and the Exchange Services (OA, OWA, ActiveSync, etc.) were running like a charm.

 

Could it be that the customers 6500 Switch (VSS-Mode) have some problems when learning the same MAC on multiple VLAN's on different ports during Failover? Does any body know about a known bug/issue pointing in that direction when using the VSS technology in combination with F5 BIGIP?

 

Thanks for any kind of advice Regards Thrillseeker

 

application delivery
availability
cisco
LTM

6 Replies

Sort By
  • Thrillseeker_12's avatar
    Thrillseeker_12
    Icon for Cirrus rankCirrus
    Sep 05, 2015

    I checked /var/log/ltm and /var/log/messages as well but no indication for an ip-address conflict. By the way this customer is using v11.6 HF4.

     

    anybody else? Thanks Thrillseeker

     

  • Patrik_Jonsson's avatar
    Patrik_Jonsson
    Icon for MVP rankMVP
    Sep 07, 2015

    I think now we've entered the long shots domain, and I'd advise you to contact F5 support.

     

    That said:

     

    • What did the switch forwarding tables say?
    • Have you tried going back to GARP?
    • Did you run a tcpdump (on both the F5's and the sending device on the external VLAN?)
    • Did you have PL on all interfaces?

    If you find the issue, please update this thread. :)

     

    /Patrik

     

  • Thrillseeker_12's avatar
    Thrillseeker_12
    Icon for Cirrus rankCirrus
    Sep 15, 2015

    Hi all,

     

    Just for your info. We recently disabled MAC-Masquerading and went back to default GARP.

     

    No ping losses anymore => problem solved! ;-) So I assume that there is an Issue with VSS Cisco and Cisco's CEF Feature when using F5 MAC-Masquerading. The VSS-Cluster has an uptime of over 3 years...;-) Looks like a VSS Bug. The network team will analyze the issue in more detail.

     

    Thanks a gain for your answers Regards thrillseeker