VMMark starting error

[psb1963]

I went to start the VMMark test for the first time yesterday.  Since this is my first time running VMMark I was just running a single tile from the prime client.  When I kicked it off I got the error below.  I tried both a 64 bit and 32 bit JRE, but got the same error.  The prime client is running Windows 2008 R2 64 bit. Any ideas would be appreciated.

C:\VMmark2\tools>java -Xms512m -Xmx1024m -jar C:\STAF\services\stax\STAXMon.jar

Exception in thread "main" java.lang.UnsatisfiedLinkError: C:\STAF\bin\JSTAF.dll: Can't load AMD 64-

it .dll on a IA 32-bit platform

        at java.lang.ClassLoader$NativeLibrary.load(Native Method)

        at java.lang.ClassLoader.loadLibrary0(Unknown Source)

        at java.lang.ClassLoader.loadLibrary(Unknown Source)

        at java.lang.Runtime.loadLibrary0(Unknown Source)

        at java.lang.System.loadLibrary(Unknown Source)

        at com.ibm.staf.STAFHandle.<clinit>(STAFHandle.java:306)

        at com.ibm.staf.service.stax.STAXMonitorUtil.getNewSTAFHandle(STAXMonitorUtil.java:53)

        at com.ibm.staf.service.stax.STAXMonitor.<init>(STAXMonitor.java:524)

        at com.ibm.staf.service.stax.STAXMonitor.main(STAXMonitor.java:5623)

[psb1963]

James, I appreciate the help, that definitely got me farther, but I'm still failing.  I have attached the latest results.  would you also be able to look at my VMMARK2.CONFIG and tell me if anything in that file could be messing me up?  BTW, is there any better guide to what all the different sections of that config file is for?

Thanks,

[jamesz08]

Since your network is very restricted you might be hitting the issue described here:

Troubleshooting: Slow startup with PowerCLI 4.0 U1 - VMware PowerCLI Blog - VMware Blogs

Try the fix described and see if it helps.  If you still fail then increase the RMQ_MinInitDelay to 90 or 120 in your VMMARK2.CONFIG

[psb1963]

Well James, this seems to have gotten me farther because it does show the PowerCLI connecting.  Now it tells me it can't ping my workload VMs. It says to verify that STAF is running, which it is, and I can ping the machines from a command line, any suggestions?

Thanks

[jamesz08]

Which workload can it not ping?  Mailserver requires adding STAFProc.exe to the firewall exception list.  If you did this before promoting to DC you need to do it again as dcpromo resets the list.

The command that is used is:

staf <hostname> ping ping

[psb1963]

The Workloads not pining are all the DS2, the Standby, and the OlioWeb.  If I do staf mailserver0 ping ping, I get PONG back; if I do staf ds2wedba0 ping ping, I get an exception error.  I will try your recommendation about the firewall, because I definitely set the firewall before promoting mailserver0 to DC.  I also will go into all the workloads and validate that STAF is running.

Thanks

[psb1963]

Fixed my ping problem and got past the VMMark SetUp phase. Now I'm failing on the Wkld SetUp.  Reporting some kind of issue with vMotion and LUNs.  I don't think I set anything up in the .CONFIG file for LUNS, but not sure what should have been in there.  Also, I have VMs in the cluster besides the workload VMs, will that impact the test, I noticed they should when VMMARK was taking inventory.  I am including the last set of results, I noticed a lot more files this time.

Thanks, Paul

[jamesz08]

The LUNS should be set to the names of a spare shared datastore that you will use as the deploy / svmotion targets.  You can use the same datastore. 

You are also missing the template.  You need to clone your standby VM to a template, the name should be set to what "Deploy/Templates" is in the VMMARK2.CONFIG.

It appears that your DS2DB0 vm has a disk on a local datastore (datastore2).  You should move it to a shared datastore.

Extra VMs are OK as long as you do not plan to publish the results. 

[psb1963]

Getting further, When I ran it yesterday morning I got another error pointing to SVMotion-Targetx-Poll.txt.  I had missed the note in the benchmarking guide stating (and in your reply) that I had to have a SPARE shared Datastore.  This is going to be a VSAN test, so I only had 1 shared Datastore.  I spent most of the day getting another device set up so I can have a second shared Datatstore.  When I ran it again, I got stopped on the same error as the first time, but the confusing thing is when I open the SvMotion-target0-Poll.txt file, I find not errors, here is what I get:

Checking Standby 0: OK: Not Found on TargetLUN

TDsame:0

TDdiffer:1

RETURNSTRING = ALLOK:LUNFOUND

on ALL 3 Hosts in the cluster and 1 VMs are on different LUNs

Check LUN completed successfully

By the way, when I ran it earlier, the values for TDsame and TDdiffer were reversed.

Any help would be appreciated, I am trying to get 1 successful short test run completed by tomorrow, after which I lose access to this equipment and will have to move to new equipment in 3-4 weeks.  My goal was to have my primeclient and workload VMs set up here, so I can move it all to the new equipment and start test immediately,

Also, I forgot to mention that the ping problem I had last week was due to STAFProc not starting on my SUSE based VMs, I had added the line trust level 5 machine 192.168.*.* to the usr/local/staf/bin/STAF.cfg file as directed, but it turned out it the file had to read trust machine 192.168.*.* level 5 otherwise the service wouldn't start.

Thanks, Paul

[jamesz08]

There is no error in what you posted so you will need to upload the results folder, it must be someplace else.

[psb1963]

Thanks for getting back so quickly, here are all the result files.

, Paul

[jamesz08]

Hmm, for some reason your STAX_Job log is missing.  Can you find the latest log files @ C:\STAF\data\STAF\service\log\MACHINE\<hostname>\GLOBAL and post it here?  They wil be named STAX_Job_X.log and STAX_Job_X_User.log

[psb1963]

here you go

Thanks

[jamesz08]

These look old, the last entry seems to be for a test where powercli wasn't initialized, but that problem was solved a while ago.

Is your client disk full?

[psb1963]

James, these were the latest STAX job log files and the client drive has plenty of room; HOWEVER, I realized that those 2 log files where time stamped from when I tried to make another run this morning, only to realize someone else was messing with the storage I had my Deploy SVMotion spare storage on; so I had to kill the job shortly after starting.  The results folder from earlier was the run I tried last night that had failed.  I am attaching both the results and STAX job logs from a clean run I just made, which had the same results as last night.

Thanks

[jamesz08]

OK, for some reason the retrieval of the OS Customization never completes. Deploy-OScustomization0-Poll.txt.

Can you try manually retrieving the OS Customization to make sure it is there?

Launch PowerCli then report back what is returned:

connect-viserver vc01

Get-OsCustomizationSpec Standbycust

[psb1963]

here are the results:

Get-OSCustomizationSpec : 2/9/2016 11:09:42 AM    Get-OSCustomizationSpec

  Could not find Customization Specification with name 'Standbycust'.

At line:1 char:24

+ Get-OSCustomizationSpec <<<<  Standbycust

    + CategoryInfo          : ObjectNotFound: (Standbycust:String) [Get-OSCust

   omizationSpec], VimException

    + FullyQualifiedErrorId : Common_CommonUtil_FilterCollection_ObjectNotFoun

   d,VMware.VimAutomation.ViCore.Cmdlets.Commands.GetOSCustomizationSpec

Is there somewhere in the VMMARK2.CONFIG file that I need to enter the name Standby0 that I missed?

Thanks

[jamesz08]

This is for the OS Customization spec.  This is the Deploy section in the VMMARK2.CONFIG.

pg 56 of the Benchmarking Guide describes creating this under the heading:

"Prepare the vSphere vCenter Server for the Deploy Infrastructure Operations"

Follow the steps listed.  You will need to add the IP you specify in the customization spec to your client hosts file with the hostname deployvm1

[psb1963]

What if your deploy template is built on Windows 2008? there are no system downloads for windows 2008 since they are already in the OS.

Just curious, for an official run is there a way to skip the Deploy steps?

Thanks, Paul

[jamesz08]

Since 2008 is not supported for this operation it was never tested, I am not really sure what would happen.  You would still need to create an OS customization spec as that is independent of sysprep.

You can skip deploy by changing Infrastructure operations in the INFRASTRUCTURELIST variable in VMMARK2.CONFIG. Just uncomment it and remove Deploy from the list. 

[psb1963]

OK, this morning is my last chance to get a run through before tearing it down till we get new equipment next month, I will be retaining my separate prime client and exporting my workload VMs for later use.  I tried commenting out the lines in the  "Deploy/Template" section and the line "INFRASTRUCTURELIST= "Deploy Vmotion SVmotion"" but I still get " No Deployment Template Passed". I also had tried commenting out all the lines in the "SVmotion/TargetLUN" section but that was giving me a NoNUMSVmotionLUN Passed".  I'm not as interested in running the Deploy or VMmotion, as getting the workload score.  When running this on our final test equipment the goal will be to only have a single VSAN Datastore.

Thanks