High consumption of CPU after upgraded to 11.6.0

Also, there is a huge structural change in v11.6.0, and you will have 50% less of the number of tmm threads running than before the upgrade. The tmm processes are now each pinned to a specific CPU core, and half of the cores are dedicated to admin tasks only. See the Release Notes.

I have the same issue on a 4200v with LTM and ASM, 11.6 hf3. Whenever I turn ASM policy on for a VS cpu usage jumps to 100%. Top shows mysqld and learn_manage as most consuming processes and management plane cores to 30% each. However I don't have a cluster to see if failover occurs, LTM doesn't seem to be impacted by this high cpu usage.

Hum.. I can see anything about Policy learning... If I active a policy learning i see the cpu jumps to 100% and the top proccess is learn_manage so I turned off the learning and everything sounds good... BUT I DON'T want to leave this off... something wrong here!! SURE!

When our pair did a fail over and looking for logs I discovery that failover was because the disc latency... anything about 10seg to response.... and saw in someplace about this version was changed anything about how write logs... so.... I m sure about this change on the logs has something with policy learning... maybe a exccees of disc IO... i dont know

I have the same issue on a set of 4000's running LTM, ASM, APM, AVR, and AAM in a failover HA configuration. top shows that mysqld, tmm, and learn_manager are using the majority of the CPU. At times I see mysqld using 95%+. I put in a ticket with F5 today. Another issue I'm having (which may be related), is that pulling up illegal requests (Security->Event Logs->Application->Requests) has gone from 30 seconds for the page to load to 5 minutes plus since we upgraded from 11.4 to 11.6 HF3. I have another ticket in for that, but I'm wondering if the issues are related.

Thanks Shane!! Please, share to us if you have any answer from F5.

I quoted the F5 response at the bottom of this post. Since we haven't seen any latency through the F5, I'm going to consider this a non-issue for now. I'm still waiting to hear back as to why it is taking so long to pull up the Security event logs:

From F5 support:

"Hi Shane,

As we discussed, the graphs under Statistics > Performance > System CPU Usage > View Detailed Graph, will show you CPU utilization per core an what you'll see is that just Core 7 is running high and thats probably due to learning mode being enabled.

The other issue is a memory leak ramp caused by monpd which you already knew about.

So basically as long as Core 7 doesn't stay pegged at 100% for hours then the box is running within spec."

11.6HF4 has been released a few days ago. Does it solve your problem?

lol I did the upgrade yesterday, for now it appears that the consumption is the same... lets wait for the next days to know if it solve the crash dump