Masking of CPU to vcenter

Similar Questions

• Masking of CPU vs. VCA

  What is the diff between masking of EVC and CPU? Aren't the same thing? The doc mentions how CVS is used in clusters and you can use a selected base line but generally it obscures features just CPU which is about masking of cpu.

  Until AMD and Intel streamline their feature sets so vmotion works out of the box. This isn't a viable solution. I don't think that anyone here does not dispute this fact.

  I gave information on the masking of the AMD and Intel cpu for example to show how powerful it is, and why CVS must be considered on the masking of CPU unless you need something special.

• vMotion mask CPU

  I'm trying to vMotion to an Intel Xeon Quad Core 2.5 GHz E5420 processor to an Intel Xeon Dual Core 2.33GHz 5140.

  I activated the No. run protection and memory virtualization technology intel in the bios on both hosts.

  From what I can see the vMotion compatibility groups B and C.  For B & lt;-& gt; C vMotion, he told me to apply SSE4.1 mask (http://kb.vmware.com/selfservice/microsites/search.do?language=en_US & cmd = displayKC & externalId = 1991)

  The SSE4.1 mask is stored in ecx and the mask is-

  ---

  -

  ---

  -

  ---

  0--- -

  ---

  -

  ---

  -

  ---

  -

  ---

  I did this and tried to vMotion to another host and get the following message is displayed.  Don't know what else to try.

  It's in the DRS cluster settings.

  Click with the right button on the cluster and go on change settings...

  You will see a frame for VMware EVC.

  If I remember correctly, this can be live.  Virtual machines must be turned off.  Try it. It won't let you turn on VCA he wants power off virtual machines.

  VCA is basically the masking of CPU for you.

  I currently have 3 Nehalem servers (Xeon 5500) and a Xeon 5400 host in my cluster. VCA works perfectly.

• HA Cluster and vMotion CPU compatibility

  Will be the following servers and work of the processor all in a HA Cluster with ESXi embedded 4.1 with vCenter 4.1 vSphere?

  Reference Dell 2950 III Xeon E5410 @ 2.33 GHz

  Reference Dell R610 Xeon X 5550 @ 2.67 GHz

  Reference Dell R710 Xeon X 5650 @ 2.67 GHz

  Or will I have to activate the VCA Mode on the Cluster?

  Is there a chart that shows you what processor works in a HA Cluster and vMotion?

  You need a CPU mask on the VM vCenter migrating to the EVC cluster and adding one would require a reboot.  You can do instead.

  (1) create new cluster with appropriate VCA setting.

  (2) (2) migrer migrate virtual machines from one host, put it in maintenance mode and then move it to the new cluster.

  (3) stop vCenter.

  (4) connect to the host that was running with the vSphere client vCenter.  Unregister the virtual machine.

  (5) connect to the host in the new cluster, to save the VM vCenter and turn on.

  Dave

  VMware communities user moderator

  Now available - vSphere Quick Start Guide

  You have a system or a PCI with VMDirectPath?  Submit your specifications to Officieux VMDirectPath HCL.

• Configure several CPU with HA

  I am runing 3 server cluster, with ESX4.0, with 24 GB of memory each and Dual CPU

  vCenter Server Standard 4

  case 1

  case 1

  vSphere 4 Enterprise (1-6 cores per processor)

  6 processors

  6 processors

  I can not configure more than 1 CPU if I activate HA, is there a limitation with license

  There is a limit of 1.0 FT know.

  See also: VMware FT

  André

• Activation of VCA mode for cluster production

  Hello

  I have the following configuration:

  • Existing Cluster_A with ESX 3 hosts running on the processor Intel Xeon E5540 (Nahelem)
  • 30 VM runs in a production environment
  • 2 new ESX hosts running on the processor Intel Xeon E5649 (Westmere)

  2 new guests are not added in vCenter again, but the goal is to allow the addition of these 2 new hosts in Cluster_A.

  I understand that, in order to have all 5 hosts in the same cluster, EVC mode must be activated with the highest level being "Nahelem" (lowest common denominator).

  From a theoretical point of view, masking any CPU to apply will be identical to the current CPU presented the virtual running machines, settings allowing CVS using the same architecture CPU should not cause conflict or problem.

  Therefore, I would ask you confimation that activation VCA and Nahelem for Cluster_A upgrade will not cause downtime or impact (e.g. BSOD) to the virtual machines.

  Thank you

  Kenny

  Therefore, I would ask you confimation that activation VCA and Nahelem for Cluster_A upgrade will not cause downtime or impact (e.g. BSOD) to the virtual machines.
  Kenny

  Hello

  Is 100% safe operation (done many, many times the performance of clusters). No side effects :-)

  1. create a TEST VM
  2. Add the new server ESX to vCenter
  3. put in maintenace mode
  4. Add to the cluster
  5. allow to LCA
  6. exit maintenance ESX
  7. try to vMotion between the old and the new node in a cluster
  8. If vMotion won't work, says André, you should turn off the virtual machine and run it.
  See you soon
  Artur
  

  Please, do not forget the points of call of the "useful" or "correct" answers

• EVC cluster creation / migration hosts him / minimal disruption VM

  I am reshape my somewhat environment and need a bit of advice.  Currently, I have 2 ESX hosts in a non - EVC cluster and built a new blade server.  The new blade and 2 hosts will work together in a cluster active CVS.  I know that I have to run my EVC cluster "Intel Xeon Core 2" mode.  All 3 hosts will support this mode.

  The News / 3rd cluster is sitting in his own vCenter, a VCA-activated cluster (in mode above).  Has no running VMS, it is already configured with network access and storage that matches that of the 2 existing/production hosts, which are currently managed by another instance of vCenter (but I am moving them out of this instance of vCenter in this new EVC-activated, to join the new blade.

  All hosts running ESX 3.5 U4.  vCenter is 2.5 U4.

  BIOS on guests must be modified to memory XD in so that they can be used in an EVC cluster, so I know I need to stop at SOME point to get them in the cluster active CVS.

  I have an idea on how to address these requirements, with a big goal is NO interruption of service for running virtual machines.  VMs can be migrated to a new cluster, EVC-activated, coming from an old cluster VCA not compatible?  Most all current VMs based on Intel some running already in reference "Intel Xeon Core 2" EVC mode or I'm just kidding me and must arrest all of them at some point after they "seized" a VCA-activated cluster?

  How would you do to this migration between vCenters and Clusters?  I think I can use 'Disconnect' the first vcenter (to at least avoid having to go into Maint Mode and then 'delete' of the first cluster/vCenter database)... also, I'm looking for the most logical get "here" to "there".  Thank you.

  I see that the problem of different vCenters/Clusters is resolved.

  What you need now is to know if you can VMotion between two ESXs with different processors. Before existing VCA, I have had... and had NOT a few difficulties doing this. It is not a problem when DIFFERENT pCPUs have the SAME GAMES of INSTRUCTIONS (SSE3, SSSE3, SSSE4 and so on...) Even if the sets are different, I was able to mix the MASK of CPU management options, which are by VM (thing that I didn't like... There should be an option to change this on a same CLUSTER basis or an ESX!).

  I recommend you to test for differences between the sets of instructions of the Jepp of two ESX with this tool, I think it is the best out there:

  http://www.run-virtual.com/?page_id=155

  If the sets of instructions are different, you have a problem, because you have to find the bits change in the MASK of CPU options in each virtual machine... and who needs a reboot, too. I think they will be different, because more recent pCPUs with FlexMigration (VCA compatible Intel processors), simply because they are new they surely have a new instruction set (perhaps SSSE4)

  Good luck!

• confusion for rdy with esxtop %

  Dear all,

  I have a virtual machine with 8vCPU, I wanted to see his rdy time % I saw this machine evolves in 2.1% to 4.3%, I read the same values can be verified using vSphere client, please find attached photo showing the status of 8 vCPU ready. But seeing SSH, I couldn't see a value between 2.1% and 4.3%, why isn't a SINGLE value when there is 8vCPU as shown in the attached photo. In customer vphere, I am able to see the 8 values, while one to match with the inSSH seen one

  I read that the acceptable range is up to 5%, in my case according to what I've seen both with SSH and vSphere client seems ok. But I need to understand what I've observed, if correct, this status is ok and I need to consider something for my vCPU as reduction VM etc.

  Thank you

  I want to check is to look at its usage statistics real CPU in vCenter over 1 day and 1 week, or even a month if your stats go back that long and see what that's mean the CPU usage is.  In my opinion, who will decide if you can retrieve this VM resources.  If the virtual machine CPU usage is maxing out at 30 percent and most of the time on average very low you can easily retrieve resources.  So everything using the RDY to see if the host takes too long to schedule time for your virtual machine, looking at the use of the processor for the virtual computer will tell you used how heavily the virtual machine's resources are used.

• Meaning of lines hostCPUID, userCPUID, and guestCPUID in the vmware.log file

  I'm trying to get a solid knowledge about the exact meaning of the hostCPUID, userCPUID and guestCPUID entries in the vmware.log file.

  I'm well versed in masking of CPU and I know how to change masks with cpuid. < number >. < enter > = 'XXXXXXX '.

  I guess the 'host' in hostCPUID is the bits real raw CPUID reported to ESXi by the host CPU.

  What is less clear, exactly what are the values 'guest' and 'user '. Based on the name, guestCPUID looks like the bits of functionality exposed to the virtual machine. However, I don't see how the guestCPUID is established, in view of the hostCPUID and my own cpuid. < num >. < Register > masking. Maybe there is some other implicit mask takes into account as well?

  And finally "userCPUID"is completely baffling to me." Maybe something to do with the CPUID features in a mode not ring-0?

  Any clarification would be helpful. Many hours of research on Google are not getting obvious answers.

  Thank you

  Matt

  MattPietrek wrote:

  I'm trying to get a solid knowledge about the exact meaning of the hostCPUID, userCPUID and guestCPUID entries in the vmware.log file.

  I have poured into the masking of CPU and I know how to change masks with cpuid. . = "XXXXXXX".

  I guess the 'host' in hostCPUID is the bits real raw CPUID reported to ESXi by the host CPU.

  What is less clear, exactly what are the values 'guest' and 'user '. Based on the name, guestCPUID looks like the bits of functionality exposed to the virtual machine. However, I don't see how the guestCPUID is established, in view of the hostCPUID and my own cpuid. . masking. Maybe there is some other implicit mask takes into account as well?

  And finally "userCPUID"is completely baffling to me." Maybe something to do with the CPUID features in a mode not ring-0?

  Any clarification would be helpful. Many hours of research on Google are not getting obvious answers.

  Thank you

  Matt

  You are basically correct.  GuestCPUID represents the bits of functionality exposed to the virtual machine.  In addition to your mask, there are masks applied depending on the features supported by the virtual hardware.

  UserCPUID, this is what will be visible to code ringtone-3 comments running in native mode if you use the binary translation.  With binary translation, usually only ring-0 code (or IOPL-3) is subject to the binary translation.  Most ring-3 of the code is running in native mode (in the mode that we call "direct execution.")  Before the introduction of CPUID failing, it was impossible to intercept comments the CPUID instruction execution when the guest was performed in direct execution.  Some processors support a limited ability to derogate some CPUID leaves (on a basis of registry by registry) even without intercept the CPUID instruction.  Therefore, userCPUID is based on hostCPUID, but records that can be replaced have values of guestCPUID.

  These fields are rewritten in every market, so there is little interest to change them manually.

• HP Flex10 and config recommended vDS for HA and failover?

  In the last two days I have experienced connectivity issues with my VM and I think I reduced it to beacon probing when is used as a failover mechanism in my groups of port / VLAN. I guess that the broader question "is what the settings recommended for failover based on my hardware configuration"...

  My ESXi hosts run 5.0.0U1 or 5.1.0 I just walking to vCenter 5.1.0a. Hosts are BL460c G7 servers in a c7000 with ICs Flex10 chassis. Each ESXi has a profile of server blade with 4 FlexNICs: a pair for the movement of data all VM (several Tagged VLAN) that belong to a vDS. The second pair is used for all traffic management and (at the moment I have reset) is on a stand-alone vSwitch with untagged traffic.

  Uplinks for each IC go with the Avaya VSP9000 switch. The two VSP9ks are connected by a STI trunk with all the VLAN tagged.

  Initially, I discovered the source of my traffic has dropped with the ARP table of a computer virtual pretend flip flop between the local binding of 10 GB on an Avaya and the link of the IST for the other Avaya (i.e. the uplink of the other IC module was used), which can cause a small temporary loop.

  No matter what method of load balancing I configure in the PG (stick to those supported by VC) and regarless if I have the two dvUplinks marked as Active or a marked as Active and the other as the day before, I see occasional about-face when beacon probing is used.

  I've set up virtual networks in Virtual Connect to use Smart Link, so I hope that Link Status should only be enough.

  Now while I still don't entirely understand beacon probe (I read about it on a few posts), I was hoping that it could provide a little more resilience despite having Smart Link set up on the side of VC.

  So for those of you who have the same hardware configuration, how you have configured your?

  Long term, I would like to bring to the management of virtual machines inside the vDS, mainly to have more flexibility in the management of bandwidth to have a slice of the 10 GB uplink established Virtual Connect level. I tried to put the management in the vDS but failed miserably and I think it's down to a scenario of chickens and eggs and I get the order to execute orders wrong... I'll look in that, once my configuration problem above fundamental network has been sorted.

  Thank you

  regarding overhead... NIOC is... good... but the general rule is that we release the core esx... so if we use more features... it will finally be an overhead projector... and if physical hardware does the same... thing... so it's better... This is why vmware has now VAAI... SRIOV... CPU/Memory unloading... If someone does not have the blades... and then if they have 10gig NIOC is only option...

  http://pibytes.WordPress.com/2013/01/12/multi-NIC-VMotion-speed-and-performance-in-vSphere-5-x/

  Management and vMotion are on the same VLAN (but different IP addresses) - it is recommended to use a VLAN from a safety perspective... and in production, this is how we do it. they can share the same NIC... no problem... but it should be labeled. 2 natachasery with 2gbe is good - I have recently done a comparative analysis of the speed of vmotion in the blade, you can see more info in my blog - http://pibytes.wordpress.com/2013/01/12/multi-nic-vmotion-speed-and-performance-in-vsphere-5-x/

  So a few questions:

  • I understand that you recommend keeping management / management of vMotion in a pair separated from flexNICs way to not give an additional charge within VMware.

  It is not mandatory... we have to use separate NICs... its all based on the customer's environment. But in separate VLANS... of couse VST marking consumes cycles CPU... we can ignore... the world... This is ignored. in case, you can use 2 natachasery with 2gbe to combine the mgmt/vmotion.

  • Y at - it an advantage of splitting management and vMotion in different VLAN but keep in the same uplink (so the VLAN tagging)

  security wise and best practices in production, we use separate VLANS, and we can share the same NIC...... There are some use cases as... that if you have 100 guests... and there may be several vmotions happening at the same time... in this case, we use dedicated pSwitch so that VMOtion traffic will not FLOOD the main switch. so here we use natachasery dedicated... in a small environment and a well balanced and if not on subscription of cluster CPU/RAM, then there is very little happening VMotion... If we can share the same natachasery and same pswitch.

  • Would we be better to use two different pairs of FlexNICs, one for management and one for vMotion using 3 pairs of FlexNICs in total (1 for all data VLAN, one not marked for management and a third no marked for vMotion). If so, I guess you'd give different bandwidths in VC to keep vMotion, nice and fast?

  According to my bench marking... what I did... If we give... 2GbE for vmotion, going too fast... and once again, as I said above... We can combine... or just give 2 natachasery with a bandwidth of 500 MB for mgmt and 2 gig for vmotion for VM traffic rest... * it is a good design... Here, we have isolated physically and VLAN wise as we need to isolate.

  for the low response fo the console... check the DNS resolution... and nothing to do with... the bandwidth. Check the CPU/RAM vcenter and the Vcenter CPU/RAM database...

• vMotion between Intel and AMD

  Hi all

  I have read many sites mentioned my question but I'm a little confused, so I want to ask here if someone can help me.

  I have a test lab with a ML350 G6 (with qnap 439 pro II iscsi), now I have another server and I want to do a few tests/Labs. But the problem is that the new server is a microserver HP with AMD processor. With HA activated when I try to vMotion virtual machine, I get errors on the incompatibility of the CPU. So, how to get around the CPUiD mask? I tried with the maskering official vmware KB the RDTSCP and CMPXCH16B, but it does not work.

  Is there anyone who has had the same problem? Or should I only use a cold migration?

  A big thank you to all and have a nice day

  Luke

  In fact, I know that I must or "must" use cold migration.

  Hello

  You use cold migration :-) no reverse

  But there is also an official KB of vwmare http://kb.vmware.com/kb/1993 saying and explaining how to hide and mask id CPU of derivation.

  As has been said in the previous answer, you have to have a CPU with the same provider, regardless of if you want to create the CVS base or masking of CPU by VM

  See you soon
  Artur
  

  Please, do not forget the points of call of the "useful" or "correct" answers

• Management EFXi 4.0 free version

  Hello

  Is it possible to manage virtual machines running on a server ESXi 4.0 automatically?

  I have the free version of ESXi 4.0 and a free version of vSphere. This combination works very well, but I want a solution based on a startup script, stop and backup of all my virtual machines. I tried "vSphere CLI" and 'VMware VIX', but it seems that they do not work in combination with the free version of ESXi 4.0.

  If there is no way to do it with the free version: what license should I buy?

  concerning

  Robert

  Free ESXi is limited to read-only controls.

  Less expensive license is vSphere Essentials - 1 k$. You will have licenses for 3 hosts dual CPU and vCenter Foundation.

  And of course, you will be able to start and stop virtual machines with scripts.

  ---

  MCSA, MCTS, VCP, VMware vExpert 2009

  http://blog.vadmin.ru

• Need to change the settings vmx CPU-mask on the fly

  

  In a migration scenario that I need to change the CPU mask without a chance to turn off virtual machines for the changes.
  Is it possible to do without crashing it?

  I do not have the exact settings, but I can give them later.
  At the moment all the tips are welcome

  It's advanced vCenter (set to false).
  Administration--> vCenter Server Settings--> Advanced settings

  Here's the brutal approach;

  config.migrate.test.CpuCompatibleWithHost
  -With this setting, all the stable related to CPU compatibility throughout the vCenter Server is disabled.  If you have ESX/ESXi hosts on different hardware, will not be tested the CPU compatibility, and this could have implications serious impacts on virtual machines.

  But you might want to watch using the following instead.

  config.migrate.test.CpuCompatibleMonitorSupport
  -Including the value of this parameter, then 'version of the product does not support functions' errors will be removed, but other errors of CPU compatibility will be still tested for.

  config.migrate.test.CpuCompatibleError
  -The value of this parameter, then CPU compatibility warnings is still displayed in the migration wizard, but they do not block the migration.

  See you soon,.

  Jon

• CPU masking SSE4.1 - how to?

  We have two esx host running vSphere 4 processors but are in different compatibility groups, and we must operate the vMotion.  They are 5140 Intel processors and Intel 5450.  According to this article: http://kb.vmware.com/selfservice/microsites/search.do?language=en_US & cmd = displayKC & externalId = 1991 - they are CPU comatibility groups B and C.  To resolve this issue at the level of the VC, it is said to apply SSE4.1 mask.  I would like to try this, but I have no idea how to go on this subject...

  I know we can do global masking by editing the file vpxd.cfg on the vCenter server, however I cannot find examples of what exactly should be changed or edited.  A few brief details are listed here: http://kb.vmware.com/selfservice/microsites/search.do?language=en_US & cmd = displayKC & externalId = 1993 #VC2xESX2x

  Are there any examples or help to edit this file to apply SSE4.1 mask would be greatly appreciated!

  Thank you!

  SSE 4.1 is a powerful multimedia instruction set, so unless you run scientific applications or treatment video you will see no difference at all.

  ---

  VMware vExpert 2009

  http://blog.vadmin.ru

• [Solved] VCenter SQL query eat CPU resources

  Does anyone else have a problem with an SQL query in the database of Vcenter eat lots of CPU resources?

  We noticed that most of the time a core processor is 100% used by the SQL Server, dig deeper, we found that it is always the same query runs over and over again:

  SELECT DISTINCT 
                           VPX_SCSI_LUN.UUID AS lunUuid, CASE VPX_HOST_BUS_ADAPTER.VPX_TYPE WHEN N'vim.host.BlockHba' THEN CAST(VPX_PSA_PATH.HOST_ID AS VARCHAR(255)) 
                           + '-' + VPX_SCSI_LUN.UUID WHEN N'vim.host.ParallelScsiHba' THEN CAST(VPX_PSA_PATH.HOST_ID AS VARCHAR(255)) + '-' + VPX_SCSI_LUN.UUID ELSE VPX_SCSI_LUN.UUID END AS scsiVolumeEntityId, 
                           VPX_DATASTORE.NAME AS datastore, VPXV_DS_LUN_INFO.DATASTORE_ID AS datastoreEntityId
  FROM            VPX_DATASTORE INNER JOIN
                           VPXV_DS_LUN_INFO INNER JOIN
                           VPX_PSA_ADAPTER INNER JOIN
                           VPX_PSA_PATH ON VPX_PSA_ADAPTER.HOST_ID = VPX_PSA_PATH.HOST_ID AND VPX_PSA_ADAPTER.KEY_VAL = VPX_PSA_PATH.ADAPTER_LINK_KEY INNER JOIN
                           VPX_PSA_DEVICE ON VPX_PSA_PATH.LUN_LINK_KEY = VPX_PSA_DEVICE.KEY_VAL AND VPX_PSA_PATH.HOST_ID = VPX_PSA_DEVICE.HOST_ID INNER JOIN
                           VPX_HOST_BUS_ADAPTER ON VPX_PSA_ADAPTER.HOST_ID = VPX_HOST_BUS_ADAPTER.HOST_ID AND VPX_PSA_ADAPTER.LINK_KEY = VPX_HOST_BUS_ADAPTER.KEY_VAL INNER JOIN
                           VPX_SCSI_LUN ON VPX_PSA_DEVICE.HOST_ID = VPX_SCSI_LUN.HOST_ID AND VPX_PSA_DEVICE.LINK_KEY = VPX_SCSI_LUN.KEY_VAL ON VPXV_DS_LUN_INFO.UUID = VPX_SCSI_LUN.UUID AND 
                           VPXV_DS_LUN_INFO.HOST_ID = VPX_PSA_PATH.HOST_ID ON VPX_DATASTORE.ID = VPXV_DS_LUN_INFO.DATASTORE_ID
  WHERE        (VPXV_DS_LUN_INFO.DATASTORE_ID IS NOT NULL)

  sp_who2
  SPID     Status     Login     HostName     BlkBy     DBName     Command     CPUTime     DiskIO     LastBatch     ProgramName     SPID     REQUESTID
  52        RUNNABLE                           NT AUTHORITY\SYSTEM     NIF-VCENTER02       .     nif-vcenter02_vcdb     SELECT               20273570     1436     10/20 12:38:26     Microsoft SQL Server JDBC Driver                   52        0

  Everyone has noticed this problem with Vcenter 5.5 Build 2183111 (update 2) and MSSQL 2012 SP2 + CU2?

  Who like a lot of information storage, perhaps associated refresh "Views of storage" in the vSphere client.

  For the mentioned tables, can you update the statistics on them and see if that helps?