Forum Discussion

Aditya_Mehra's avatar
Aditya_Mehra
Icon for Cirrus rankCirrus
Dec 18, 2017

Clients unable to connect during business hours

Hi All,

 

So I have a Virtual Server set up in an LTM. Virtual server listening on port 6000, using usual TCP profile and source based persistence, and 2 pool members on 6000 port. The URL works fine, but during peak hours (when network traffic is high) the clients complain unable to login. When clients try to connect directly via the pool members it connects fine, that is if we bypass the Virtual Server. Do not see any logs or errors on the LTM logs.

 

If anyone can share any insights on this.

 

Thanks, Aditya

 

application delivery
LTM
  • rsacheen's avatar
    rsacheen
    Icon for Nimbostratus rankNimbostratus
    Dec 18, 2017

    I guess the use of source based persistence is playing a role here. Might be your VS is overloaded maintaining persistence information during peak hours. How about trying without persistence?

     

  • Aditya_Mehra's avatar
    Aditya_Mehra
    Icon for Cirrus rankCirrus
    Dec 18, 2017

    Hi Rsacheen, Yes there is Persistence used. Can you please explain " VS is overloaded maintaining persistence information during peak hours"

     

    How would we know that it is overloaded? Thanks, Aditya

     

  • Aditya_Mehra's avatar
    Aditya_Mehra
    Icon for Cirrus rankCirrus
    Dec 18, 2017

    Hi Jaikumar, We are using the standard VS setup.

     

    Automap is used, Protocol - tcp, Protocol Profile (Client) - tcp , Protocol Profile (Server) - using client profile, and source persistence.

     

    Thanks, Aditya

     

  • jaikumar_f5's avatar
    jaikumar_f5
    Icon for Noctilucent rankNoctilucent
    Dec 18, 2017

    Do you find any logs in /var/log/ltm something like below,

    01010201:2: Inet port exhaustion on
  • jaikumar_f5's avatar
    jaikumar_f5
    Icon for Noctilucent rankNoctilucent
    Dec 18, 2017

    You may need to take tcpdump and ssldump during the event to know more on the issue.

     

  • Aditya_Mehra's avatar
    Aditya_Mehra
    Icon for Cirrus rankCirrus
    Dec 18, 2017

    Thanks Jaikumar , right now do not see this issue as its intermittent.

     

    Will take a tcp and ssl capture to see further if anyone reports.

     

    do you have any insight on the above mentioned VS overload maintaining persistance table.

     

    Thanks, Aditya

     

  • jaikumar_f5's avatar
    jaikumar_f5
    Icon for Noctilucent rankNoctilucent
    Dec 19, 2017

    Hi Adi,

     

    I'm not aware of any issues that could arise from persistence table.

     

  • NAG_65570's avatar
    NAG_65570
    Historic F5 Account
    Dec 26, 2017

    You can use following Plan of action to capture traffic specific to client experiencing the issue and find the cause for the issue::

     

    1) tmsh modify /sys db tm.rstcause.log value enable 2) tmsh modify /sys db tm.rstcause.pkt value enable

     

    3) Capture the packets on LTM using the following commands.

     

    tcpdump -ni 0.0:nnnp -s0 -w /var/tmp/.pcap

     

    4) capture the packets on the client host experiencing the issue using wireshark.

     

    NOTE: both step 3 and 4 should be performed simultaneously while client experiencing the issue

     

    5) terminate the logs by using following commands. tmsh modify /sys db tm.rstcause.log value disable tmsh modify /sys db tm.rstcause.pkt value disable

     

    By Analysing the Captures in wire-shark and LTM logs, you should be able to find the root cause for the issue