2 Replies Latest reply on Jul 12, 2013 6:32 PM by bobby2004

    ESXi 5.1 - Auto Deploy Stateless Caching Question on HP Servers

    aaronwsmith Hot Shot

      Working on setting up Auto Deploy to use stateless caching for our HP ProLiant servers with Smart Array Controllers (ESXi uses the hpsa driver to talk to this controller.)  In the host profile, enabled stateless caching and set "Arguments for first disk" = "hpsa."

       

      While this seems to discover the local disk properly, I find applying the host profile takes a very long time and watching vmkernel.log the host is scanning all the SAN attached VMFS datastores for ESXi installations (or so it seems, see attached picture.)

       

      I also tried "esx,local,hpsa" and got the same behavior.

       

      I'm curious what other people are using for enabling stateless caching on ProLiant servers?

       

      Thanks!

        • 1. Re: ESXi 5.1 - Auto Deploy Stateless Caching Question on HP Servers
          aaronwsmith Hot Shot

          Made some progress here.

           

          I figured out (I believe) all the options that seem to be valid for setting the first disk argument to stateless cache on an HP ProLiant server with a Smart Array controller and a single RAID1 logical volume:

           

          hpsa

          hp

          logical volume

           

          All of them result in the same behavior so I went with hpsa to ensure I always target the logical volume that's created through the HP Smart Array controller.

           

          Regarding the scanning of SAN volumes for ESXi installs, we have a support ticket open with VMware to understand this behavior because we're finding when a host profile with stateless caching is enabled on an Auto Deploy'ed VMhost with a large number of SAN volumes (in our case, 32 SAN volumes, each 1TB - 2TB in size), during bootup, the VMhost fails to completely apply the profile.  When we scale back to about 15 SAN volumes, host profile with stateless caching applies to the VMhost every time.  The issue seems to be linked to the number of SAN volumes ESXi 5.1 must scan as part of stateless caching when booted via Auto Deploy with a host profile.

           

          I'll update this topic once I have more information.

          • 2. Re: ESXi 5.1 - Auto Deploy Stateless Caching Question on HP Servers
            bobby2004 Lurker

            Hi,

             

            I am working on same config (deploying esxi 5.1 using autodeploy stateless caching) and had issues. If these are HP proliant blade server then these are arguments are fine "esx,local,hpsa"...the only issue is that if you are reusing old server which already had esx install and.even if you delete the raid (which doesn't delete the metadata completely - even erase utility from HP does not work) and recreate it ...the esxi installer reads disk properties and finds metadata on disk and considers it as snapshot volume..you can run command to verify that disk will be there but would need resignaturing. this is a bug (worked with vmware on this). Workable workaround is to reset all settings including erasing disk in bios....go back and reset date/time (reset sets it to default and if time is wrong auto deploy will not add host to vcenter), change boot order (pxe first, disk 2nd), go to controller and delete and then recreate array. After all these steps..autodeploy works like charm and configures boot banks and local datastore (reconfirm caching by booting to disk). let me know if you need any further help.