Forum Discussion

Andreia_Hora_14's avatar
Andreia_Hora_14
Icon for Nimbostratus rankNimbostratus
Mar 14, 2018

F5 LTM VE - Health Monitor Problems After Update 12.1.2 to 12.1.3.2

Hello, everyone.

 

I upgraded the F5 BIGIP VE version 12.1.2 to version 12.1.3.2. However, I had to rollback this update due to unexpected pool behavior. A few pools with Health Monitor TCP, for example, had members with OK status. The members that were supposed to be ONLINE were ONLINE and those that should be OFFLINE were OFFLINE, however, most pools were with all members with "ONLINE" status and members who were supposed to be offline had " unkwown "(active). The pools themselves, some presented online status and others as "unknown." Looking at members of these pools, members who were supposed to be offline appeared as "unknown" and the Health Monitor icon appeared as a ticking clock. The pools that have fqdn members also have "unknown" status. We have several pools with different types of health monitor and all have exhibited this strange behavior.

 

 

In the ltm log, in / var / log, the pools had one with normal behavior. For example, members with status "unkown (active)" did not appear in the log as online. Before updating the BIGIP, I put its status as OFFLINE. However, soon after I tried the update with BIGIP in ONLINE mode and I had the same answer. I tried restarting the bigd, too, and yet the strange scenario remained. The applications were lost because the members of the pools that were supposed to be offline started receiving requests. I know this behavior of unkwon stays for a few seconds after the boot, but I waited for 10 minutes and the environment did not normalize.

 

Could you help?

 

2 Replies

  • Sounds like you should submit a support case. Be sure to take a QKView and UCS backup of the affected BIGIP on the upgraded code, prior to any rollback. If there is a production issue, you should probably utilize 'switchboot' to return to the previous working logical volume and configuration.

     

  • Surgeon's avatar
    Surgeon
    Ret. Employee

    As polskikrol explained, it is better to open a support case. Please be aware that F5 Technical Support Team will ask you to reproduce the behaviour, take qkview and capture file at issue time.

     

    If you had the same issue on both HA peers then you can go with standby box. This should not impact any service.