none
SCVMM 2012 RC

    Question

  • RC?  More like Beta 2.  This is not a question, just my feedback on SCVMM 2012.

    I am recent Hyper V/SCVMM convert from VMware.  I really have had few complainst until now.  So here is my impression of SCVMM 2012 RC

     

    1. Upgrade over SCVMM 2008 R2 SP1 with all known patches failed with a error about a "null" value.  Looked in the setupwizard log and found the error but it was vauge at best.  Attempts to re-install gave the same error.  Even though it said rolling back, my 2008 R2 install was toast.
    2. New, clean, install went well enough, if I used a new database, so my NULL value problem must be in teh old database (SQL 2008 R2 fully patched).
    3. Adding hosts....failed until I removed the old agent, so much for easy upgrade.
    4. Speed of the tool.  There is where the "beta" part shines bright.  This thing is HORRIBLY slow.  I am running it on a server with 8gigs of RAM and a quad core, same as with 2008 R2.  The database is tiny (80megs?)  so there is no real I/O.  The CPU's are all over the place with an average on all 4 at 30%, even after 24 hours.  The RAM is barely being used.  I only have 12 hosts and about 180 VM's.  Creating a new VM on a cluster takes FOREVER on a 6 node cluster while it scans the hosts.  My hardware is not slow, Dell R610 with dual 6 core Xeons at 2.93ghz, 96gigs of RAM per host, 10gig LAN, dual 10gig iSCSI with jubo frames etc, all running through Nexus 5548's.  Works way faster in 2008 R2.
    5. After clicking through...15 min of dialog boxes to place a test VM on a cluster, some of the nodes have Zero stars because there is not enough disk space????  Some are fine, all attempt default to the C drive of the host.  Even though on each host in Hyper V manager its pointed to a CSV SAN volume, and in the properties of the host in VMM I choose one of the CSV as the default.  I cant choose those hosts, it wont let me.  If I pick one of the nodes that has stars it to wants to install it on the C drive??  I can override but there goes a few min of my life waiting for the dialog box to show up and be usable....again the speed.
    6. My SAN (Equallogic) does not support SMI-S yet.  So seeing the storage in the GUI is not great. In VMM 2008 I could click on the host and choose the "Network and Storage" tab to see the CSV volumes and their space etc.  I can see this info in VMM 2012 if I rght click on a host and go to properties....after I waaaaaaiiiiiiiit for that to come up.  Maybe I am missing something but its not easy to see what is going on with my volumes, or harder than in 2008 R2.
    7. Dynamic optmization is cool.  I have only stoodup Hyper V and VMM 2008 R2 in recent months and I have held off adding my SCOM server into the mix because I wanted to do it after SCVMM 2012 was out.
    8. The other users of VMM in my group, missed the Window at the bottom that shows the VM running, you could see where a VM was....like as it was booting up etc.  Why was this pulled in VMM 2012 or did we just miss where it is??

    Anyhow my fault for going the RC route, should have waited.  I am back on 2008 R2, at least the un-install/re-install was painless.  I wont be jumping on this new version ANYTIME soon, unless I read of some massive improvments in speed and those funky cluster storage problems go away.

    Time to get SCOM plugged into VMM 2008 R2.


    Thanks, -Lindy
    Friday, October 07, 2011 5:36 PM

Answers

  • I was not trying to make any comment.  Simply make sure that it is clear what an RC is.  And do direct you to the proper location for your feedback so the team is guaranteed to see it.

    In my experience an RC is really no more than a very late stage beta.

    Does this mean that an RC has no bugs?  Nope, they still have bugs. 

    They have just pretty much stabilized on what is there.  You will very rarely see new features between and RC and release.  But there are always bug fixes.  Just no new features or feature changes.

    I work with pre-release software all the time and because of that I am sure that I have differnet expectations.  At the same time, I have a low tolerance for bugs.  Especially in what I consider common tasks.

    I have seen some software applications release just as many RC builds as betas.  It really all depends on the company.

     


    Brian Ehlert (hopefully you have found this useful) http://ITProctology.blogspot.com
    Learn. Apply. Repeat.
    • Marked as answer by cwlindy Friday, October 07, 2011 8:39 PM
    Friday, October 07, 2011 8:31 PM

All replies

  • You do realize that you installed "Beta" software over your production system?

    RC is "release candidate"  - this means that: "this is not the final release but it feature complete and pretty stable.  There still might be issues"

    If you wanted a really good upgrade experience then wait until SCVMM 2012 is in "RTM" or the final release.  That is when all the testing will be finished regarding the upgrade process.

    Your other feedback points are good.

    Please feel free to give your feedback to the SCVMM folks directly through connect.microsoft.com

    Look for "system center in the datacenter" and you will find SCVMM 2012.

     


    Brian Ehlert (hopefully you have found this useful) http://ITProctology.blogspot.com
    Learn. Apply. Repeat.
    Friday, October 07, 2011 6:36 PM
  • Ummm yeah I figured I would get some comments like yours.  I stated that it was not a question and that is was feedback.  I also said it was my fault for going the RC route.  As far as my production system...yeah my backups worked great thanks for asking.  Even with out a backup, just like VMware, Hyper V kept runnig, so did the VM's and in the case of Hyper V where clusters are not part of VMM/vShpere they kept running as well.  Only 4 of us even knew I went down this path and it was quickly rolled back.

     

    To me RC is NOT beta.  Beta is bug filled this is what it might or might not have in it, check it out...use at your own risk. 

     

    RC is...this is a feature finished, bug squashed as far as we know,  Candidate for Release, please check it out .....use it at you own risk. 

     

    Most RC's I have used in the past feel no differen than the RTM when it is all said and done.  I am sure there are some differences but nothing I can usually tell.  Windows 7 is a great example.  If this is what the RTM of what VMM 2012 is going to be like....wow, Microsoft needs a RC2 at least.

     

     


    Thanks, -Lindy
    • Edited by cwlindy Friday, October 07, 2011 7:43 PM
    Friday, October 07, 2011 7:42 PM
  • I was not trying to make any comment.  Simply make sure that it is clear what an RC is.  And do direct you to the proper location for your feedback so the team is guaranteed to see it.

    In my experience an RC is really no more than a very late stage beta.

    Does this mean that an RC has no bugs?  Nope, they still have bugs. 

    They have just pretty much stabilized on what is there.  You will very rarely see new features between and RC and release.  But there are always bug fixes.  Just no new features or feature changes.

    I work with pre-release software all the time and because of that I am sure that I have differnet expectations.  At the same time, I have a low tolerance for bugs.  Especially in what I consider common tasks.

    I have seen some software applications release just as many RC builds as betas.  It really all depends on the company.

     


    Brian Ehlert (hopefully you have found this useful) http://ITProctology.blogspot.com
    Learn. Apply. Repeat.
    • Marked as answer by cwlindy Friday, October 07, 2011 8:39 PM
    Friday, October 07, 2011 8:31 PM
  • Thank you for your feedback Lindy, it is very valuable.

    The team would be very interested in your upgrade issues.  If you could submit those logs and description at connect.microsoft.com, it would be very helpful indeed.

    Your zero star storage issue suggests to me that you were trying to deploy a non-HA VM onto your cluster. VMM2012 now supports deploying non-HA VMs onto local storage on cluster nodes, but perhaps the UI did not make it clear enough that that was the underlying cause of your storage issues.

    I'm glad you like Dynamic Optimization!

    Thanks, Hilton

    Saturday, October 08, 2011 4:57 AM
  • Here is the error message...

     

    02:15:26:VMMPostinstallProcessor threw an exception: Threw Exception.Type: System.ArgumentNullException, Exception.Message: Value cannot be null.

    02:15:26:StackTrace:   at Microsoft.VirtualManager.Setup.VirtualMachineManagerHelpers.CopyApplicationFrameworkFiles()

       at Microsoft.VirtualManager.Setup.InstallItemCustomDelegates.PangaeaServerPostinstallProcessor()

    02:15:26:ProcessInstalls: Running the PostProcessDelegate returned false.

    02:15:26:ProcessInstalls: Running the PostProcessDelegate for PangaeaServer failed.... This is a fatal item.  Setting rollback.

    02:15:26:SetProgressScreen: FinishMinorStep.

    02:15:26:ProcessInstalls: Rollback is set and we are not doing an uninstall so we will stop processing installs

    02:15:26:****************************************************************

    02:15:26:****Starting*RollBack*******************************************

    02:15:26:****************************************************************

    02:15:26:SetProgressScreen: StartMinorStep.

     

    For the storage issues.  I currently have a 6 node cluster I use for Windows Server VM's.  All of nodes are identical, Dell R610's, with 2 dual port 10gig cards, 2 NIC ports dedicated to iSCSI on their own VLAN.  My SAN is a group of Equallogic storage units (4 of them) all on the same 10gig VLAN as the servers.

     

    I have 9 CSV volumes assigned to this cluster and I have about 60 VM's spread across those 6 nodes, and 9 CSV volumes.

     

    If I create a new VM in VMM 2008 R2 I get to the point where I want to place it, choose that Host Group that has the cluster, it scans those 6 hosts (quickly) suggests one, then I get an option to place it on storage, it picks my first CSV always, which is a CSV on SATA I use for DEV servers, I can drop down and change it to another CSV if I want.

     

    In 2012 it takes literally 5-10 min to scan the hosts?  After it does only 3 of 6 nodes have stars.  The other 3 have a red circle for storage, saying something about not enough storage, if I try to choose one of them it throws an error.  If I pick one of the three that have stars they all default to the local hard drive of that server (80gig mirror on the Dell Server).  My assumption is that it is trying the local drive of the others and maybe they dont have enough space on the local drive?  The test deploy was a 40gig 2008 R2 template (fixed disk) so maybe its close on all of them.  I am not sure why it is even considering local storage?  On the hosts that I can deploy I am able to choose other storage, by browsing to the C drive and picking the folder with the CSV's in it and picking the correct one....that process is very slow to respond each step of the way in the GUI.

    2012 is gone now so I cant really tell you what the errors were now.  I also looked at host properties under the "placement section" and I could see the CSV's and chose the top one "CSV-DEV1" and clicked OK on all of them.  Also at the local level of each host in Hyper V manager they are all pointed to the same location for the default storage option.  It could be that I did not know how to set the default option, but I did look around quite a bit. 

    Also like I mentioned above it is hard to see the status of my CSV's, as in their current space avalibility etc.  You need that information when placing new VM's.  It is easier in 2008 R2, before I place a VM I go to the host/top of cluster and see which server has the most VM's and then take a look at the volumes to see what has the most space....then when placing the VM I have idea of where it should land.

    There other things I like about 2012, the ability to have it auto generate static MAC's via a template vs clicking the "generate" button in 2008 R2 after the deployment.

     


    Thanks, -Lindy
    • Edited by cwlindy Saturday, October 08, 2011 3:29 PM
    Saturday, October 08, 2011 3:26 PM
  • Hello cwlindy,

     

    you can only place a vm on a csv if you set the machine as high available in the hardware configuration tab. Per default it is set to normal availability. Also you need to set the volume c:\ in the hardware tab of the nodes as available for placement. Otherwise you can't place vm's on csv.

     

    greetings stefan

    Tuesday, October 11, 2011 8:56 AM
  • Exactly the same issues as Lindy described. (null valuse error, clean install, adding host error, agent winrm errors...)
    After the third step, I restored the system to its previous state (aka SCVMM 2008R2 wSP1).

    Apparently Microsoft must release RC2, 3, 4, before Final release.

     

    Wednesday, November 02, 2011 9:45 AM
  • Seems like as good a place as any to leave feedback. I installed the RC on a 6 node cluster 2008 R2 SP1 cluster. Immediately after the agent started installing on the hosts, two of them failed with a BSOD and restarted bringing a sizable chunk of VMs offline. This also left some of the CSVs in a weird locked state so they were stuck in the failed state until the hosts were rebooted a second time.

    The computer has rebooted from a bugcheck. The bugcheck was: 0x0000007e (0xffffffffc0000005, 0xfffff800016d35c0, 0xfffff88002ea70f8, 0xfffff88002ea6950).

    After installation, all but one host were stuck in a pending state. After trying to get them working for a few hours I rolled back to SCVMM 2008 R2, at which point it became apparent that Winrm was completely screwed. Two days of searching for a solution, I found the following two commands had to be issued to get the agents talking again;

    winrm quickconfig

    winrm set winrm/config/service @{EnableCompatibilityHttpListener="true"}

    The second command enables Winrm to use port 80 as well as the new port, otherwise SCVMM cannot communicate with the agents.

    Im only posting this as feedback, I know all the caveates about not using RC in production. Hopefully

     

    Wednesday, November 16, 2011 8:30 AM
  • Hi Mrphil45

    Same error (other bluescreen) after start the agents on my cluster... The agents works o.k. in non-clustered server.

     

    Thursday, December 01, 2011 1:27 PM
  • Just to clarify on this, for VMM 2008, when the host is a node in cluster, VMM will handle the paths for the virtual machines. So, there is no way to setup that in the Placement Tab in the  Host Properties.

    http://technet.microsoft.com/en-us/library/ee236431.aspx

    noteNote
    When you add a host cluster, you do not specify default virtual machine paths, as you would for a stand-alone host. For a host cluster, VMM automatically manages the paths that are available for virtual machines based on the shared storage that is available in storage groups.

    Hello cwlindy,

    you can only place a vm on a csv if you set the machine as high available in the hardware configuration tab. Per default it is set to normal availability. Also you need to set the volume c:\ in the hardware tab of the nodes as available for placement. Otherwise you can't place vm's on csv.

    greetings stefan


    Tuesday, February 07, 2012 7:24 PM
  • I have installed SCVMM 2012 RC on one of our nodes for evaluation and so far I've been quite impressed although I haven't had the chance to really get into the nitty gritty yet. One thing I've noticed that is a bit disconcerting is that there is no Hosts view. In the 2008 version, there is a Hosts view where I can see the status of each of my hosts. The one that interests me most is being able to see when a host in is maintenance mode, for example. In 2012, I haven't yet found out where I can see that information. The only way I have of knowing whether a host actually went into maintenance mode when I told it to was to look in the Jobs view. Is there a way to get a Hosts view in the new version?
    • Edited by VernCook Wednesday, February 22, 2012 3:37 PM
    Wednesday, February 22, 2012 2:28 PM