Forum Discussion

Rakesh_Reddy_Ba's avatar
Rakesh_Reddy_Ba
Icon for Nimbostratus rankNimbostratus
Feb 23, 2016

Virtual server is frequently getting disconnected with ERR_CONNECTION_RESET on BIG-IP 11.5.1 , Please Help!!

Hi ,

 

We have a JBOSS based enterprise application hosted on F5 BIG-IP 11.5.1 with HF10 VE.

 

we have 2 virtual servers (https based) configured with same ip on different ports 8888 & 8089.

 

we are terminating the ssl connectivity on F5 LB using clientssl & serverssl ( we are using self signed certs for test purpose).

 

when i access my application with ,sometimes it is loading my login page but sometimes i am receiving connection reset. if i login to my big-ip UI and remove http profie or add http profile or sometimes simply click on update on my virtual server configuration,i am getting the connection again.

 

it will work for some time and again getting reset.

 

The same application use to work nicely with out any issues with BIG-IP 11.31 VE.since we need TLSV1.2 suport we moved to the newer version.

 

I have tried several troubleshooting information available in devcentral but could not break through. Please help me in resolving this issue. Thanks Rakesh

 

3 Replies

  • Josiah_39459's avatar
    Josiah_39459
    Historic F5 Account
    Is anything logged in /var/log/ltm when you get the reset? If you capture both front and back end, is the reset coming from the backend server or from the vip?
  • If you're out of ideas, you can try implementing a few changes I have listed below. The app is already broken and unavailable anyway, right :)?

     

    Personally, I'd start by enforcing the use of TLSv1.0 (seen as SSLv3.1 in SSLdump utility) on the serverssl profile you use for this particular Virtual Server. I have seen a number of issues with incapable end-servers that do not respond/downgrade if SSL handshake's ClientHello from BigIP comes in with a version that is not supported. A SSL handshake problem on the serverside is one of the possible reasons for receiving a TCP/RST on the clientside.

     

    Another idea. Since your issue is occuring sporadically, it could be that your problem is isolated to one of the end-server listener ports, and not both of them. Try disabling port TCP 8888 listener Pool Member(s), and see if the issue can be reproduced. Repeat vise-versa with the other Pool Pembers.

     

    Just a few ideas on how to continue.

     

    Good luck,

     

  • Thanks for the response,I have found that the issue was with the network related ,we are frequently losing connectivity in our Lab. I configured the same in another network and everything working as expected.