Forum Discussion

Chris_FP's avatar
Chris_FP
Icon for Cirrus rankCirrus
May 26, 2015

HSL Publisher does not recover when the pool goes down and then up

We have a set of iRules which produce logging to allow the back-end team to diagnose issues and due to the sheer volume of log entries these are not stored locally but via a dedicated "HSL VLAN" on a 20gig vPC link.

 

setup is:- The pool has one pool member, the log destination is set to Remote-HSL, the aforementioned pool attached and using TCP (UDP isn't an option). Finally the log publisher itself has the log destination as its attached destination.

 

The pool has a simple tcp-half-open monitor to confirm that it is up.

 

What we have noticed is that should the pool go down and then back up again, the logging from the HSL commands in the iRules don't get to the destination any longer. The only way to kick it back into life is to attach the local-syslog destination (the inbuilt one), hit update and it kicks back into life.

 

We need to use a publisher rather than hard coding an IP as the name of the publisher to use is held in a DG so that the logging iRule Proc can be used in multiple environments. Plus we can send a particular VS iRule logging to a different destination should we choose due to the granular nature of the design.

 

Before I raise a ticket with F5, wondered if anybody else had come across this issue?

 

(same issue exists in 11.4.1 and 11.5.1 and 11.5.2)

 

1 Reply

  • can you open a support case and check whether it is ID503101?

     

    ID503101 HSL: investigate culling connections to pool members marked down by monitors.