Feed for content matching tag 'ft'

Survey on VMware Fault Tolerance (FT)

bchanana — Mon, 15 Mar 2010 19:10:05 GMT

I'm in search of current and future use of FT. Any feedback you can provide would be very valuable to our ongoing product efforts to develop and enhance VMware Fault Tolerance and VMware HA. Please use the link below to respond to the questions in the survey or forward to the appropriate individuals in your organization that can respond.

http://www.surveymethods.com/EndUser.aspx?AD89E5F7ABEEFDF8

thanks!

NB: I'm cross-posting from the "Availability:HA & FT" forum in case you've already seen this message once before.

VMware Fault Tolerance Survey and Feedback

bchanana — Tue, 09 Mar 2010 01:24:35 GMT

Increasing timeout of re-enabling secondary VM

slavdok — Fri, 05 Mar 2010 20:47:11 GMT

Hi

I have 2 ESX hosts, 1 FT-enabled VM on shared storage.

VM is running on host1. When host1 is disconnected, the VM fails over to host2 (this works fine).
When the host1 is connected back to the network, ESX tries startup secondary VM on this host now (primary is successfully running on host2 now).

My problem is that host1 does not have access to the shared storage for some time after it is reconnected to the network. This is a limitation of my infrastructure and cannot be changed.

Host1 tries to access the shared storage for some time, fails, and gives up, leaving the secondary VM powered off. Even when shared storage becomes available, the VM remains powered off.

I need to be able to set this VM/Host to continously try bringing up the secondary VM. I know it's possible cause i heard of some parameter that increases the timeout for trying to re-enable FT.

Please help

Thank you.

system requirements vs. fault tolerance for vCenter Server

jodell1 — Sat, 27 Feb 2010 01:55:48 GMT

Background: I have a small vSphere cluster that I am currently testing configurations on. The cluster includes between 3 - 6 hosts. My vCenter server resides as a VM in the cluster. My initial configuration included HA and FT for the VM runing vCenter Server. FT requires a single virtual CPU, however the system requirements for vCenter server is 2 CPUs. In my current iteration, the vCenter VM has two vCPUs. HA is available, as in my failure tests, works very well (I kill the host, thus the VM and vCenter die, one of the remaining hosts adds the now dead vCenter VM to its inventory, restarts is and in a few minutes everything is back to normal). My goal though is to not ever (unless absolutely necessary) have my vCenter server go down, thus FT.

Question: With such a small cluster 3-6 hosts, can I get away with a single-vCPU, self-contained (database on the same server using SQL bundled with vCenter install) vCenter server? (since the single-vCPU will allow me to activate FT on the VM)

Some Technical Info: 3 x Dell PowerEdge T300's, (2 - 3 Dell PowerEdge 2950's will be added later)1 x EnhanceTech UltraStor 16 IP-4 (adding a second soon) vCenter Server, vConverter Running on Server 2003 R2, all hosts are running ESXi and all vSphere components are at Update 1 build level.

Thanks in advance.

FT failover with loss of iSCSI SAN connectivity

ADVM — Fri, 29 Jan 2010 18:17:17 GMT

We are running two identical ESXi 4.0 update 1 hosts and vCenter Server. On our hosts we have redundant network connections and redundant iSCSI connections. We currently have one guest OS set up with FT. In testing the failover, we physically unplugged both connections to our SAN from the host running the primary instance of our guest OS. In doing so, we expected to see FT take over and fail our guest over the the secondary host. We expect this to happen based on the following excerpt from the "Protecting Mission Critical Workloads with VMWare Fault Tolerance" whitepaper (http://www.vmware.com/files/pdf/resources/ft_virtualization_wp.pdf):

"Component Failures

ESX already protects against component failures, such as the
failure of an individual SCSI host bus adapter (HBA) or network
interface card. The storage multipathing functionality in ESX
automatically fails over to a new HBA if the currently active HBA
stops working. Similarly, the NIC teaming functionality in ESX
automatically uses another NIC if the active NIC fails. These component-
level failovers happen automatically, and no FT failover
is required. On the other hand, if all HBAs fail on the primary
host but HBAs are still functioning on the secondary host, the
VMware Fault Tolerance causes a full failover to the secondary
virtual machine, so the workload can continue."

The behavior we observe on the guest OS is that it appears to continue running and accepting commands, but some applications hang and the experience is slow/unresponsive. Once we reconnect to our SAN, it doesn't appear as if any instructions to the machine were lost (as in normal functionality is restored) or even that there was any data loss, but this is not the behaviour we were expecting.

Can anyone shed some light on how this is "supposed" to work or is it possible we've mis-configured somewhere? As a side note, when we force a failover from vCenter, it works beautifully.

obscure FT guest failure after hardware malfunction

friedchicken — Mon, 22 Feb 2010 21:16:37 GMT

Hi there,

an interesting one happened to me today. i have a 2 node cluster that i've just set up FT for my first VM. it powered up and went through the motions of starting the FT. unfortunatly it never got as far as getting the secondary fully started before the primary server hardware died.

now the vm is stuffed - it has gone to a disconnected status. the options to power on/off are all grayed out. i can't remove it from the inventory either. i tried re-adding the guest from the datastore browser and it won't let me.

any ideas what to try? it may be a few days before i can get the primary server fixed and i could really do with the vm back up and running!

thanks in advance.

Secondary VM in a fault tolerance configuration reports using a higher percentage of memory than primary

tdupont — Thu, 21 Jan 2010 15:04:21 GMT

I recently created a fault tolerant domain controller for our VM environment and I noticed that the secondary VM is reporting that it is using 87% of configured memory even though it is using the same MB of memory as the primary.

I have the VM configured to use 2GB of memory. The Primary is reporting that it is using 755MB and the secondary is reporting that it is using 768MB of memory but when you look at the column that displays guest memory percentage the primary is reporting 5% usage while the secondard is reporting 87% usage.

Is this normal behavior? I would think that they would be configured and reporting nearly identical conditions.

Converting from PVSCSI back to LSI

tdupont — Mon, 18 Jan 2010 19:07:49 GMT

Recently we created our templates to use PVSCSI adapters not realizing that you can not use this adapter with Fault Tolerance. We built a new domain controller using the template as a base and now we would like to make it FT.

Every time we try and switch the adapter it will blue screen. I have tried the normal method of adding a hard disk to SCSI port 1:0 and making the new SCSI controller LSI while leaving the original SCSI controller PVSCSI. We booted into Windows so that it would load the LSI drivers. We rebooted, changed the old hard disks to be on the new controller and rebooted again. Once we did that, the system blue screens. We can change it back to PVSCSI and it will boot into Windows.

The VM is running Windows 2008 R2 and I tried to repair the boot process using the Windows 2008 R2 DVD with similar results.

I tried to P2V the existing VM in the hopes that during the process it would convert the PVSCSI adapter to LSI for me, but I get an error during the conversion process that it could not write more than one active partition to the partition table. I think this may be due Windows 2008 R2 having a hidden boot partition on the primary hard disk.

Any assistance would be greatly appreciated.

vmotion, ha and ft licensing

DavidDS — Wed, 20 Jan 2010 12:38:34 GMT

I have a doubt, if I need FT or HA is necesary license vmotion? How work the licensing of vmotion, HA and FT? Thanks,

esxcfg-portgroup-mgmt.pl

lamw — Wed, 20 Jan 2010 05:36:47 GMT

Author
Description
Category
Features
Requirements
Version Support
Sample Execution

Author

William Lam

Description

This script was motivated by VMTN community member and blogger Jason Boche who posed a question on the VMware forums regarding a Scripted solution for enabling Management Network on a portgroup.

The script supports not only enabling and disabling the management network on a supported portgroup type but also Fault Tolerance Logging and vMotion.

Features

Enable and Disable Fault Tolerance Logging, vMotion and Management Traffic

Requirements

ESXi 3.5 or 4.0
vCLI OR VMware vMA 4.0

Version Support

Supports ESX(i) (licensed version only)

Usage

[vi-admin@scofield scripts]$ ./esxcfg-portgroup-mgmt.pl
Required command option 'operation' not specified.
Required command option 'portgroup' not specified.
Required command option 'portgroup_type' not specified.

Synopsis: ./esxcfg-portgroup-mgmt.pl OPTIONS


Command-specific options:
   --operation (required)
      Operation to perform [enable|disasble]
   --portgroup (required)
      Name of portgroup
   --portgroup_type (required)
      Type of portgroup [vmotion|ft|mgmt]

Common VI options:
   --config (variable VI_CONFIG)
      Location of the VI Perl configuration file
   --credstore (variable VI_CREDSTORE)
      Name of the credential store file defaults to <HOME>/.vmware/credstore/vicredentials.xml on Linux and <APPDATA>/VMware/credstore/vicredentials.xml on Windows
   --encoding (variable VI_ENCODING, default 'utf8')
      Encoding: utf8, cp936 (Simplified Chinese), iso-8859-1 (German), shiftjis (Japanese)
   --help
      Display usage information for the script
   --passthroughauth (variable VI_PASSTHROUGHAUTH)
      Attempt to use pass-through authentication
   --passthroughauthpackage (variable VI_PASSTHROUGHAUTHPACKAGE, default 'Negotiate')
      Pass-through authentication negotiation package
   --password (variable VI_PASSWORD)
      Password
   --portnumber (variable VI_PORTNUMBER)
      Port used to connect to server
   --protocol (variable VI_PROTOCOL, default 'https')
      Protocol used to connect to server
   --savesessionfile (variable VI_SAVESESSIONFILE)
      File to save session ID/cookie to utilize
   --server (variable VI_SERVER, default 'localhost')
      VI server to connect to. Required if url is not present
   --servicepath (variable VI_SERVICEPATH, default '/sdk/webService')
      Service path used to connect to server
   --sessionfile (variable VI_SESSIONFILE)
      File containing session ID/cookie to utilize
   --url (variable VI_URL)
      VI SDK URL to connect to. Required if server is not present
   --username (variable VI_USERNAME)
      Username
   --verbose (variable VI_VERBOSE)
      Display additional debugging information
   --version
      Display version information for the script

Sample Execution

Enable Management Traffic

[vi-admin@scofield scripts][http://esxi4-2.primp-industries.com|http://esxi4-2.primp-industries.com]$ ./esxcfg-portgroup-mgmt.pl --server esxi4-2.primp-industries.com --operation enable --portgroup "Management Network" --portgroup_type ft

Enabling "faultToleranceLogging" on portgroup: "Management Network" ...

Enable FT

[vi-admin@scofield scripts]$ ./esxcfg-portgroup-mgmt.pl --server esxi4-2.primp-industries.com --operation enable --portgroup "Management Network" --portgroup_type ft

Enabling "faultToleranceLogging" on portgroup: "Management Network" ...

Enable vMotion

[vi-admin@scofield scripts]$ ./esxcfg-portgroup-mgmt.pl --server esxi4-2.primp-industries.com --username root --operation enable --portgroup "Management Network" --portgroup_type vmotion

Enabling "vmotion" on portgroup: "Management Network" ...

Disable Management Traffic

[vi-admin@scofield scripts][http://esxi4-2.primp-industries.com|http://esxi4-2.primp-industries.com]$ ./esxcfg-portgroup-mgmt.pl --server esxi4-2.primp-industries.com --operation disable --portgroup "Management Network" --portgroup_type mgmt

Disabling "management" on portgroup: "Management Network" ...

Disable FT

[vi-admin@scofield scripts][http://esxi4-2.primp-industries.com|http://esxi4-2.primp-industries.com]$ ./esxcfg-portgroup-mgmt.pl --server esxi4-2.primp-industries.com --operation disable --portgroup "Management Network" --portgroup_type ft

Disabling "faultToleranceLogging" on portgroup: "Management Network" ...

Disable vMotion

[vi-admin@scofield scripts][http://esxi4-2.primp-industries.com|http://esxi4-2.primp-industries.com]$ ./esxcfg-portgroup-mgmt.pl --server esxi4-2.primp-industries.com --operation disable --portgroup "Management Network" --portgroup_type vmotion

Disabling "vmotion" on portgroup: "Management Network" ...

VMware FT

AndreTheGiant — Sun, 25 Oct 2009 10:33:08 GMT

FT Information

http://www.vmware.com/products/fault-tolerance/
Meet the Engineer Series: VMware Fault Tolerance
A Summary of FT Related Documents
http://www.vmware.com/files/pdf/perf-vsphere-fault_tolerance.pdf - VMware vSphere™ 4 Fault Tolerance: Architecture and Performance
http://www.vmware.com/files/pdf/resources/ft_virtualization_wp.pdf - Protecting Mission-Critical Workloads with VMware Fault Tolerance
http://tech.philipsellers.com/2008/09/20/introduction-to-vmware-ft-fault-tolerence/
http://kb.vmware.com/kb/1010601 - Understanding VMware Fault Tolerance
http://kb.vmware.com/kb/1013428 - VMware Fault Tolerance FAQ
FT not supported on Nehalem systems with hyperthreading enabled?

About vLockStep:
http://software.intel.com/en-us/blogs/2009/05/28/instruction-level-lock-step-parallelism-on-desert-islands/
http://mordtech.com/tag/vlockstep/

Demo of VMware's vSphere Fault Tolerance (FT)
{youtube}http://www.youtube.com/watch?v=NCMMwGC0hD8{youtube}

FT requirements

http://www.ntpro.nl/blog/archives/1090-Fault-Tolerance-Checklist.html

Only some CPU can be used: http://kb.vmware.com/kb/1008027 - Processors and guest operating systems that support VMware Fault Tolerance
Hardware Virtualization (HV) must be enabled in the BIOS for each host in the cluster (but this is also required for EVC and x64 VMs)
The ESX hosts that run the primary and secondary Fault Tolerance nodes, must be running the same build of ESX.
All ESX hosts used by VMware Fault Tolerance must be members of a VMware High Availability (HA) cluster. VMware HA must be enabled for VMware Fault Tolerance to function.
Networking must be configured properly to ensure proper VMware Fault Tolerance functionality.
In some cases the VM must be powered off to enable FT (for some guest type and to convert vmdk from zeroedthick to eagerzeroedthick )

To check compatibility there is also a VMware utility:
Site Survey - http://www.vmware.com/download/shared_utilities.html

FT 1.0 limitations

Only 1 vCPU for each VM protected by FT
Require compatible CPU (http://kb.vmware.com/kb/1008027)
No vmxnet3 support (http://kb.vmware.com/kb/1013757)
No snapshot support (and no VCB and other snap-based backup program)
No physical RDM (Need clearification on FT requirement)
Do not use both ESX and ESXi hosts in clusters with fault tolerant virtual machines (http://kb.vmware.com/kb/1013637)
VM memory will be reserved to the max value
Only eagerzeroedthick disks are supported (VMDK virtual disk type)
Seems that Linux guest with VMI are not supported for FT

FT 1.0 best practices

Use dedicated 1Gb (better if is 10Gbs) link for FT logging
Use SVMotion to converter from thin to thick, or use dastore browser/inflate (but in this case the VM must be powered off)
Use CPU with similar speed
Do not use CPU power management
With Nehalem CPU, HT could be enable without problem (and is also suggested by VMware)
Enable FT with the VM powered off

FT vs VM HA or MSCS

FT does not protect against OS or application failures, only from ESX host failures.

See also:
Clarificaiton on Fault Tolerance
http://blog.scottlowe.org/2009/10/30/vmware-ha-vmware-ft-and-os-clustering/

VMware FT vs Marathon eveRun

http://www.yellow-bricks.com/2008/09/24/marathon-haft-vs-vmware-haft/
http://www.mikedipetrillo.com/mikedvirtualization/2008/09/marathon-and-vmware-ft.html
http://www.mikedipetrillo.com/mikedvirtualization/2008/09/marathon-everrun-vs-vmware-ha-another-mess.html
http://itknowledgeexchange.techtarget.com/server-virtualization/vmware-defends-its-upcoming-fault-tolerance-feature/
http://channelvirtualization.wordpress.com/2008/09/18/marathon-on-vmwares-ft-sort-of-fault-tolerance/

Enable/Disable FT with script

Could be useful to implement some backup solution on FT protected VM.
on-demand FT, can this be automated?
ftCLI.pl - VMware Fault Tolerant Managment

VMware FT (Fault Tolerance): appunti

Feddar — Wed, 24 Feb 2010 10:44:59 GMT

Pubblichiamo qui velocemente un paio di appunti che svilupperemo in seguito (forse):

VMware FT non supporta gli snapshot! Per cui occhio che non puoi utilizzare vDR (VMware Data Recovery) a meno di impiegare "accrocchi" (maggiori dettagli in seguito)

L'articolo prosegue a questo indirizzo: VMware FT (Fault Tolerance): appunti

FT thick-eagerzeroed disk

pdarcy — Tue, 22 Dec 2009 15:32:43 GMT

I've enabled FT on a virtual machine which did not have a thickeagerzeroed disk on it. The process is going through but its taking a very long time and its a production system that is down. I thought that it would be completed in about an hour but it now taken 2.5 hours for 39%.

Is it possible to cancel the process and if so would the os and vmdk be in tact?

Thanks

Understanding VMware Fault Tolerance benefits and requirements

karavinds1 — Tue, 15 Dec 2009 16:48:04 GMT

VMware Fault Tolerance (FT) is a new high availability feature in VMware's vSphere 4. With Fault Tolerance, your virtual guest machine runs on a primary ESX host server while the memory of that virtual machine (VM) is mirrored (using vLockstep) over to a secondary (ghost) ESX host server. If the primary ESX Server fails, the virtual machine immediately resumes operation on the secondary ESX Server with zero downtime to the VM, application or end user using that VM.

The benefits of VMware Fault Tolerance

Unlike traditional high-availability (HA) technologies, VMware's Fault Tolerance works regardless of the operating system and you aren't charged for every server that uses this HA feature. FT is based on VMware Workstation's Record/Replay features that can play back what happened in a VM.

Fault Tolerance is not used for load balancing - it is strictly for high availability of a VM in the event that an ESX Server goes down. Unlike VMware HA clusters (VMHA), with FT, the VM that was protected doesn't have to be rebooted. Thus, with FT, unlike VMHA, there is no downtime for end users.

The best thing about VMware FT is that, to enable it all you need to do is to right-click on the primary VM and enable Fault Tolerance. At that point, FT takes over, creating the secondary VM that will protect the primary VM if the ESX server running the primary VM has a failure.

The hardware requirements for VMware Fault Tolerance

VMware FT requires Intel 31xx, 33xx, 52xx, 54xx, 55xx, 74xx or AMD 13xx, 23xx, 83xx series of processors (or greater). Moreover, you can't run FT on just any vSphere-compatible server because FT uses special features of the CPU. Today, FT is also supported only on VMs that have a single CPU (no multiprocessor VMs can use FT).

High availability, in general, is an excellent part of any large scale disaster recovery strategy. With Fault Tolerance you have the added protection for smaller-scale disasters such as the loss of a single ESX host server running 50+ virtual machines.

VMware's Fault Tolerance (FT) is supported in vSphere Advanced, Enterprise and Enterprise Plus. VMware's vCenter is also a required component.

FT sizing and HA Admission Control

Igor_Nemilostivy — Tue, 01 Dec 2009 14:16:06 GMT

Hi!

Plz, specify how to make a correct sizing FT.
For example, there are two virtual machines. Then, FT takes 4 slots of cluster. Plus 2 slots for SecondaryVMs from the documentation:

VMware HA includes the resource usage of Fault Tolerance Secondary VMs when it performs admission control calculations. For the Host Failures Cluster Tolerates policy, a Secondary VM is assigned a slot, and for the Percentage of Cluster Resources policy, the Secondary VM's resource usage is accounted for when computing the usable capacity of the cluster.

Total of 6 solts. Right?
If each virtual machine has 4Gb of memory, total will be 24Gb, right?

And what if the HA cluster with FT consists of only TWO hosts?

--
Igor Nemilostivy