VMware View 5.0.1 and all other Version - Choppy moving windows around

[BlueCloud]

i have read and read the forums and have not yet found an answer... i have vmware view 5.0.1 in a test environment running on a single gigabit switch to a thin client...

debugging so far

changed versions of agents

upgraded esxi to 5.0 u1

upgraded view to 5.0.1

changed switch

change thin client to wyse p20 and tryed another p20

changed to a high performance workstation with view client

changed cables

used another server - both servers were quad core xeon 54xx series with dual cpu - 64GB RAM - 8x 146GB 15k RPM HDD RAID 10 w/ 512MB Cache

turned off SSL in view

added the MKS setting of 1000 100 2

everything we do we still end up with choppy video quality even just when moving the monitor around on the screen

its laggy... lags behind the mouse about 1 second... when logged into the vmware vsphere console the console is very smooth even with 3d and aero enabled.

we are using PCoIP and not RDP as the P20 thin client does not support RDP to my knowledge... also the protocol chossen in vmware view client is PCoIP which is forced...

[dvhorvath]

I saw this happen once, and it ended up being that the parent VM hadn't been optimized for View. It was the same base image they were using for their physical machines, and had been P2V'd. They were able to minimize the effects with advanced display settings, but it wasn't fully resolved until they recreated the parent VM from a clean install in the View environment. I don't know if that relates to your situation in any way, but it seemed eerily similar, so I thought I'd share it.

Dave

[Linjo]

Hi.

Most of these that I have seen have been related to the mks.headless.pollrates.

Are you 100% sure that you have made the correct modification to the vmx-file and shutdown (not restarted) the vm, started the vm and then checked the vm-logfile that is actually reads the parameter correctly?

Sorry for being so particular but have happend to me a few times when the View-broker have had the vmx-file open at the same time and my manual changes have not been properly saved.

// Linjo

[mittim12]

Linjo, not to hijack this thread but have you seen the headless pollrates problem in hardware 7 machines or strictly hardware 8?

[whynotq]

from some one who is in exactly the same place i can empaphise with the pain you are going through. We have been fault finding and tunig this since week one of the deployment. What we have done so far is as follows:

• Created Win7 SP 1 64bit master with Win 7 optimisations from the VMware guide.
  
• created a single pool of Linked Clones, floating, with 1 CPU and 1GB RAM
  
• Users complained of "Laggy" browsing and desktop moves
  
• found articles on Teradici Server update (tried that minimal improvement)
  
• found articles on mks.poll.headlessrate (tried that minimal improvement)
  
• swithced off IE smooth Scrolling
  
• removed Resource Pool
  
• Set HP BIOS CPU powersave mode to Max Performance
  
• disabled 3D and set Max Screens and Max Resolution to "max allowed"
  
• Added 2nd CPU (Users now Happy)
  
• disabled Windows effects for displaying window content whilst moving (users happier still)
  

All of th eabove made minimal difference until 2nd CPU was added. Have tried to go back to 1 CPU but users instantly get on the phone and say it's laggy again. What we do see is the browser with active content (BBC or Reuters are used) generates 100% CPU use when it refreshes the active content areas on 1CPU VM, this is also seen in a standard image not part of the view build which indicates it may be Guest rather than View related.

Sorry it isn't definative and we still have a case open with VMware but hopefully it will save you some time.

regards

Paul

[juchestyle]

Laughing, this is not an advertisement, I am watching View Training videos and I just watched the segment about the vms and they absolutely suggest 2 cpus when you are using 3d imaging, and hardware version 8.  They also say that you can't use local mode if you have version 8, just a side note and throwing my 2 cents worth into the discussion.

[BlueCloud]

yes it seems to be grabbing the mks poll headlessrates

2012-04-06T14:24:44.939Z| vmx| DICT             scsi0.sasWWID = 50 05 05 69 cb 3f 73 80

2012-04-06T14:24:44.939Z| vmx| DICT               hostCPUID.0 = 0000000a756e65476c65746e49656e69

2012-04-06T14:24:44.939Z| vmx| DICT               hostCPUID.1 = 0001067600040800000ce3bdbfebfbff

2012-04-06T14:24:44.939Z| vmx| DICT        hostCPUID.80000001 = 00000000000000000000000120100800

2012-04-06T14:24:44.939Z| vmx| DICT              guestCPUID.0 = 0000000a756e65476c65746e49656e69

2012-04-06T14:24:44.939Z| vmx| DICT              guestCPUID.1 = 0001067600020800800822011febfbff

2012-04-06T14:24:44.939Z| vmx| DICT       guestCPUID.80000001 = 00000000000000000000000120100800

2012-04-06T14:24:44.939Z| vmx| DICT               userCPUID.0 = 0000000a756e65476c65746e49656e69

2012-04-06T14:24:44.939Z| vmx| DICT               userCPUID.1 = 0001067600040800000822011febfbff

2012-04-06T14:24:44.939Z| vmx| DICT        userCPUID.80000001 = 00000000000000000000000120100800

2012-04-06T14:24:44.939Z| vmx| DICT      evcCompatibilityMode = FALSE

2012-04-06T14:24:44.939Z| vmx| DICT  vmotion.checkpointFBSize = 36896768

2012-04-06T14:24:44.939Z| vmx| DICT                machine.id = vdi.broker.useSvi=0;vdi.broker.brokerPublicKey=MIHxMIGoBgcqhkjOOAQBMIGcAkEA/KaCzo4Syrom78z3EQ5SbbB4sF7ey80etKII864WF64B81uRpH5t9jQTxeEu0ImbzRMqzVDZkVG9xD7nN1kuFwIVAJYu3cw2nLqOuyYO5rahJtk0bjjFAkBnhHGyepz0TukaScUUfbGpqvJE8FpDTWSGkx0tFCcbnjUDC3H9c9oXkGmzLik1Yw4cIGI1TQ2iCmxBblC+eUykA0QAAkEAuDYIDdkon0CJDyk5AHQZYLm3LRN0D/8H26B8W5M61ykL0YyrHqiGW3gSuK6G2W0EK4XFcsVzzN8yFhTH9pEhiQ==;vdi.broker.brokers=veeam.vizzikinc.com ;vdi.broker.agentIdentity=agent/1b3b1e4f-19e5-4719-9916-28ab289f7715;vdi.broker.agentPrivateKeyHash=dm/1zztrd64Eyae7Spp1yg==;vdi.broker.itemDn=cn=1b3b1e4f-19e5-4719-9916-28ab289f7715,ou=servers,dc=vdi,dc=vmware,dc=int;vdi.broker.asyncSessionSeconds=150;vdi.broker.brokerPublicKeyHash=ytGO7IOE2moyGBXcIPCIog==;vdi.broker.singleuse2=0;vdi.broker.msMode=OFF;vdi.broker.agentPrivateKey=MIHGAgEAMIGoBgcqhkjOOAQBMIGcAkEA/KaCzo4Syrom78z3EQ5SbbB4sF7ey80etKII864WF64B81uRpH5t9jQTxeEu0ImbzRMqzVDZkVG9xD7nN1kuFwIVAJYu3cw2nLqOuyYO5rahJtk0bjjFAkBnhHGyepz0TukaScUUfbGpqvJE8FpDTWSGkx0tFCcbnjUDC3H9c9oXkGmzLik1Yw4cIGI1TQ2iCmxBblC+eUykBBYCFCsoYu72UNrpdV5cDYHj8rptFOOk;vdi.broker.disconnecttimeout=0;vdi.broker.singleuse=0;vdi.broker.useSysprep=0;vdi.broker.poolDn=cn=bluecloud,ou=server groups,dc=vdi,dc=vmware,dc=int

2012-04-06T14:24:44.939Z| vmx| DICT    mks.poll.headlessRates = 1000 100 2

2012-04-06T14:24:44.939Z| vmx| DICT        sched.cpu.affinity = all

2012-04-06T14:24:44.939Z| vmx| DICT           floppy0.present = FALSE

2012-04-06T14:24:44.939Z| vmx| DICT --- USER DEFAULTS ~/.vmware/config

2012-04-06T14:24:44.939Z| vmx| DICT --- HOST DEFAULTS /etc/vmware/config

2012-04-06T14:24:44.939Z| vmx| DICT                    libdir = /usr/lib/vmware

2012-04-06T14:24:44.939Z| vmx| DICT           authd.proxy.vim = vmware-hostd:hostd-vmdb

2012-04-06T14:24:44.939Z| vmx| DICT           authd.proxy.nfc = vmware-hostd:ha-nfc

2012-04-06T14:24:44.940Z| vmx| DICT        authd.proxy.nfcssl = vmware-hostd:ha-nfcssl

2012-04-06T14:24:44.940Z| vmx| DICT   authd.proxy.vpxa-nfcssl = vmware-vpxa:vpxa-nfcssl

2012-04-06T14:24:44.940Z| vmx| DICT      authd.proxy.vpxa-nfc = vmware-vpxa:vpxa-nfc

2012-04-06T14:24:44.940Z| vmx| DICT            authd.fullpath = /sbin/authd

2012-04-06T14:24:44.940Z| vmx| DICT          authd.soapServer = TRUE

2012-04-06T14:24:44.940Z| vmx| DICT vmauthd.server.alwaysProxy = TRUE

2012-04-06T14:24:44.940Z| vmx| DICT --- SITE DEFAULTS /usr/lib/vmware/config

2012-04-06T14:24:44.940Z| vmx| DICT --- GLOBAL SETTINGS /usr/lib/vmware/settings

[BlueCloud]

bump ?

if other people out there are also having this issue... please post to keep this alive.. i believe we all should get an answer regarding this issue.

[Linjo]

You should open a SR with VMware Support and get them to figure this out. Please post the SR-number here.

// Linjo

[whynotq]

my SR# 12162511504. It started life with the View support team and after 4 weeks we now have it sitting with the Guest OS team.

Paul

[mittim12]

Paul, please let us know if you reach some type of resolution and what it was.  

[BlueCloud]

the support team i got pretty much told me i need to open a case with VMware PSO... i am giving up i don't think i should or even want to sell this product to my clients... not until it either works better or something better comese along... XenDesktop 5.6 is supposed to be pretty nice... i think i'll look at possibly using this.

for now i'll stick with desktop PC's and Thin Clients connected to RD Farms with Wyse TCX Suite.

[whynotq]

I am further through the journey and to be honest i have a customer recently who deployed a Xendesktop 5.6 POC and got the same performance problems.

What we have found is when you strip away all the crap and just have a Vanilla Win 7  64Bit with SP1 and VMtools installed, open a web browser and navigate to a window with active content you still see 100%CPU use and the window occasionally jumps about. Clearly not as bad as with Domain membership, AV agent, Monitor Agent and base build software but it is still a problem.

We have now raised this with MS as we get similar CPU performance on a physical box with 1 CPU and 2 Gb RAM.

[SamuraiJack]

I just went through the troubleshooting steps to resolve this well.

Actually i tried having the users log on using RDP to see if RDP and MMR would would help but could not get the logon right. Tried using GPO with restricted groups and Remote Desktops users but still get that "allow to logon right" error even when i placed the base image in the domain and added the group to the base image Remote Desktop Users manually in Remote settings. any suggestions here would be helpful as i want to get RDP working.

Anyhoo,

Seems like View went backwards with this performance issue from 4.6 to 5 as I did not see this issue in 4.6.

The mks.poll.headlessRates = 1000 100 2 definitely helped, more than adding a second vCPU it seemed. Can this be upped any more?

HP DL460 G7 blades, 6 vitual nics, 1 GB non routed vmotion, 3GB virtual machine network and 6GB multipathing iSCSI P4500 SAN with 400GB SSD on a one partenered blade using HP VSA. 3 blades 96GB RAM and 2 socket 6 core installed on 4GB SD cards per blade. These 3 blades must hold the management servers i.e. VC, broker, VSA, FOM for P4500 and about 100 clones. All blades are optimized in the bios i.e. turbo boost, powerprofile, power regulator and thermal config all upped to max. dedicated clones have been up to 2vCPU and 4GB RAM with 8192 MB persistent disk.

Image minimzlly optimized as the users were complaing the wanted screen savers and high res pics of poochie on their desktops.

Any additional suggestions would be helpful.

Samurai Jack