DX NetOps

  • 1.  Cisco Enhanced Memory calculation

    Posted Apr 04, 2019 12:51 PM
      |   view attached

    Hi all,

     

    I'm working through some reporting discrepancies between PM and the CLI on some Cisco devices. When I manually walk a test device for cempMemPoolHCUsed and Free I see output in the ..18.5.1 and ..20.5.1 indexes that very closely resemble the CLI output of "sh mem".

     

    However, PM graphs are way off in terms of memory used, free and total.

     

    I've been looking at the metrics calculated for the Cisco Enhanced Memory Pool Vendor Cert, and am wondering if there might be an error in there. I see cempMemPoolUsed included in the calculation for memory used. Is that correct?

     

    I'll attach a screen shot of what I'm referring to.

     

    Thanks,

    David Marion



  • 2.  Re: Cisco Enhanced Memory calculation
    Best Answer

    Broadcom Employee
    Posted Apr 08, 2019 09:46 AM

    David,

     

    What that formula effectively says is:

     

    If cempMemPoolHCUsed is defined and cempMemPoolHCUsed is > 0 then use cempMemPoolHCUsed else use cempMemPoolUsed

     

    So it tries to use the 64 bit version of the OID (The HC part of it gives it away) and if it cannot, it uses the 32 bit version as a fallback.

     

    Troy



  • 3.  Re: Cisco Enhanced Memory calculation

    Posted Apr 08, 2019 02:19 PM

    Thanks Troy,

     

    I had actually found some information regarding MVEL syntax the other day after I posted this and sort of figured that was what was happening as a result of the ternary operator "?", but I appreciate you confirming it.

     

    PS - I'm about to open a new case related to this as I can now walk the test device both with the ENHANCED-MEMPOOL mib as well as the numerical OID and get values that are very close to the CLI output from "sh mem". However, I build an ad hoc report and the out put shown in that chart is very different.

     

    Thanks as always,

    David Marion