BigIP 6900 TMM Utilization

Ni Nicolas,

 

 

Each TMM should process traffic if CMP is enabled. I would guess that you're using a feature or module which prevents CMP from being enabled in your implementation. You can check SOL7751 for details:

 

 

SOL7751: Overview of Clustered Multi-Processing

 

https://support.f5.com/kb/en-us/solutions/public/7000/700/sol7751.html (Click here)

 

 

Aaron

Hi Aaron,

 

 

thanks for you answer and your link. I've just checked with bigpipe function and for each of my virtual server , CMP is enabled. I do not know exactly why these 3 last graphs are still flat as my irules are standard irules, I mean, url rewriting within http_request and response. Thanks.

 

Hello,

 

Could you try running "b virtual show all" and pipe this output to a tmp file (i.e. "b virtual show all >/tmp/vir").

 

Then look at the virtual(s) in question in the tmp file and it will have lines as follow:

 

+-> VIRTUAL HTTP_VIP SERVICE http

 

| PVA acceleration none

 

| CMP enable on none mode: all

 

 

The "mode" part is important. If it says "mode: single", this means that the particular virtual has been demoted to just one tmm (i.e. tmm1.0).

 

 

Do any of your virtual(s) show up with "mode: single"?

 

 

 

Thanks.

 

Hi,

 

 

I've launched the b virtual show command and by examining the output text file, I've only seen CMP enable on none mode: all for each of my virtual servers. In normal condition, do the 4 graphes move at the same time ? I've included a screenshot of my graphes just to show you the issue. Thanks for your help.

 

 

Regards,

 

 

Nicolas

Hello Nicolas,

 

I'm assuming that you are probably running a 9.4.x version on your 6900. FWIW, this graph doesn't exist at all in v10. I have a 9.4.5 box (single tmm) and a 10.0 box (running cmp). I see the graph in 9.4.5 but don't see it at all the 10.0 box. Since I only have one tmm in my 9.4.5 box, I can't see the problem you're talking about on my box.

 

 

There's a new dashboard in 10.0 though.

 

 

If all your virtuals are showing "mode: all" then I'd expect to see lines in all TMM's. So this means one of two things:

 

1. Either the graphs are not showing the right thing - very likely. So you'll have to work with support so they can show you in a different way that all tmm's are handling traffic (e.g. by using some CLI options)

 

OR

 

2. Your virtuals have been demoted to "single" mode (so "b virtual show all" is showing the wrong thing) and only tmm1.0 is handling traffic. Then you'll have to work with support and/or figure it out yourself by reading docs and looking at your configs to see why your virtuals might have got demoted and then work backwards to get them to "CMP mode all" again.

 

 

I know if you use global variables in irules then those virtuals get demoted.

 

 

Hope this helps.

I don't have a 6900. I have a 1600 that's running v9.4.7 and when I pass traffic through this box, I see tmm utilization graphs for all tmm's.

 

 

Are you passing traffic through the box? "b virtual show all" should show you this data at the CLI or you can use the GUI too to verify that your virtual is passing traffic.

 

 

tmm0 typically handles "other" system daemons + traffic too so the "tmm0" line could be due to other system daemons?

 

 

If you are passing traffic through your virtuals, could you post a few sample virtual configs + the attached rules to the virtuals? Also, after removing the "RULE_INIT" event from your virtual, are you able to restart your tmm? You may need to restart tmm to make sure that the change takes affect in the tmm.

 

 

Thanks.