Loadbalancing VMkernel Uplink for NFS
Hello everyone,
I'm organizing a larger facility with ethernet base storage NetApp NFS.
On the picture you can see my network design. What is the best way to configure vmkernel ports to use the two vmnic (8 and 10)?
EtherChannel / LACP between host and first switch 'instance' is not possible because the switches do not support this (HP VirtualConnect).
My fear is that, in my first setup scheduled
VMkernel Portgroup - > 1 VMkernel (192.168.1.201)-> uplink: vmnic8, vmnic10 (based on the virtual port ID)
All NFS traffic will pass through vmnic8, without worrying how can stores NFS I use. Or good vmnic8 used for the vmdk file 1, vmnic10 for a vmdk file 2, vmnic8 for the vmdk file 3 and so forth?
How many times the NFS connection road is calculated?
Will be 'Route based on the physical load of NIC' work for me?
Thank you
Marcel Mertens
Mehdi says:
Will be 'Route based on the physical load of NIC' work for me?
No, it will not help here, it might be useful with virtual switches with many VM and they could be rearranged at certain intervals if necessary.
How many times the NFS connection road is calculated?
It will be only once, according to the inner IP routing table.
VMkernel Portgroup-> VMkernel 1 (192.168.1.201)-> Uplink: vmnic8, vmnic10 (based on the virtual port ID)
All NFS traffic will pass through vmnic8, without worrying how can stores NFS I use. Or good vmnic8 used for the vmdk file 1, vmnic10 for a vmdk file 2, vmnic8 for the vmdk file 3 and so forth?
No, it will not use differnet vmnic for different vmdk files. The best way for you would be to create two NFS data warehouses and export through two addresses different IP on different IP subnets. Create two interfaces Vmkernel on the same vSwitch (with a key vmk on each subnet), and then connect them to the vmnic in settings of an active / standby. Do you use two vmnic simultaneously and also provide fault tolerance.
Tags: VMware
Similar Questions
-
Best config for uplink for VSphere
Hi all
I recently bought two Dell R710 with NIC 8 + 2 x 520 FPS + iSCSI traffic.
I designed the following config: (excluding the iSCSI)
* 1 liaison rising vmkernel-management
* 1 liaison rising vmotion vmkernel
* 2 liaison rising vmkernel-nfs
* 4 uplink for the VM network
As the storage solution is mixed with iSCSI + NFS, we wanted to provide a high resilience to the backend NFS with two uplinks dedicated.
In addition, vmotion, management and nfs services are in different VLANS in a stack of Catalyst 3750.
Management interfaces are configured in the cisco access mode, so without trunk vlan exists in any interface.
My problem is the NFS traffic is travelling through the uplink of management:-$...
Would be a good idea to join all the rising management (nfs, vmotion, mgmt) in a vswitch and configure an lacp in cisco devices to provide resilience and balance the vmkernel services? Or it is preferable to maintain the current design?
Of course, any other ideas are welcome.
Thanks in advance.
Julio
LACP would be the way to go, because you also need the traffic for vmware kernel FT.
-
Hello
I have infrastructure as follows;
2 welcome each containing 6 1Gig speed NIC each.
A NAS storage with 4 NIC
Two L2 Switches (managed HP).
Planning to run it by the best recommendations and requirements, so that there should not be a SPOF any level.
So keeping this in mind, we have designed to use ports on each server as follows:
2 for the NFS storage, 2 for Production management and 2 others for vMotion on each server.
A cable from each port configured for respective roles goes uplink switch1 and switch2, so if a switch goes down we still have the other switch support.
Separated VLAN is configured by the switch to different types of traffic.
My question as below;
Should I team two ports on each vSS.If so that what should be the parameters for NETWORK adapters for the production, storage, and vMotion network grouping (keeping in mind cables going to switch uplink separated).
Should I keep adapter in active-standby or active-active mode.
I didn't think any specific settings to link pass that one cable by a single port is to go there and I don't have the choice of the channel of the ether or LACP.
In addition, vmware license is essential and having therefore no possibility to use dynamic switches.
Consider using 5.5.0.
Also do you propose to use frames as well in the present.
Kind regards
Sushil
Hello
I suggest always you put management and vMotion on the same set of natachasery management and workloads. It makes no difference where they are subnet a perspective. I also suggest to read the following:
- The great debate VSwitch (old writings but still VERY there, let not the age fool you on this one)
- VMware Virtual networking concepts
- Networking for administrators of virtualization
Who should you get.
natachasery have no IP address in a vSphere environment, they act as a link between a physical and virtual switch. According to the way which you the trunk your VLAN Trunk ends pSwitch (external switch tagging) or the virtual switch (switch virtual tagging). Most people master their VLAN to the virtual switch.
You want something like the following:
pSwitch <->pNIC0 <->[ <->Portgroup vSwitch0] <->management (subnet1)
pSwitch <->pNIC1 <->[ <->Portgroup vSwitch0] <->vMotion (subnet2)
When switching between pNIC0 and pNIC1 management and vMotion end up on the same bear but when normally run that they remain separated. It is the recommended method. In this case you would master the VLAN to the vSwitch. I know some people who just do not use VLANs, but who use only separate subnets and it works as well.
pSwitch <->pNIC2/pNIC3 <-> <->Portgroup vSwitch1 (s) <->of workloads (subnet1)
If you use VLANs (except for vMotion) you're trunking to vSwitch1 (virtual switch tagging). If subnet1 is on the same vSwitch and the trunk is correct via pSwitch ports so he can talk to management on vSwitch0 effortlessly. Switches know how to route traffic to VLAN.
pSwitch <->pNIC4/pNIC5 <->vSwitch2 <->Portgroup <->NFS (subnet3)
Here we link pNIC4 and pNIC5 together or use them as a pair of failover for NFS on its own subnet / VLAN itself. This VLAN can end the pSwitch if you wish or terminate once again to the vSwitch.
In this configuration you have 3 VLAN and 3 subnets (subnets use by VLAN are also recommended)... for example:
VLAN100-> subnet1-> workload management->->->->->->->->->->->->->->->->
VLAN200->-> vMOtion subnet2
VLAN300->-> NFS subnet3
Let the pSwitches any 'movement' of traffic for each VLAN. You need only a routing device if you want TO cross borders VLAN and there is absolutely no need to do it in this configuration.
Best regards
Edward L. Haletky
VMware communities user moderator, VMware vExpert 2009, 2010, 2011,2012,2013,2014Author of the books ' VMWare ESX and ESXi in the business: Planning Server Virtualization Deployment, Copyright 2011 Pearson Education. ' Of VMware VSphere and Virtual Infrastructure Security: securing the virtual environment ', Copyright 2009 Pearson Education.
Virtualization and Cloud Security Analyst: The Practice of virtualization, LLC - vSphere Upgrade Saga - virtualization security Table round Podcast
-
VMKernel port done NFS use - management?
Documentation VMware - Networking ESXi 5.0 vSphere vCenter Server 5.0:
"The VMkernel TCP/IP stack handles vMotion iSCSI and NFS in the following way."
If you watch the VMkernel configuration, you can configure a port group for:
binding of vMotion, FT, lumbering, management and iSCSI ports
By the process of elimination NFS will use the virtual card for management traffic?
Right,
These boxes are available for VMware knows what vmkernal interfaces to use for the management and who to use for vMotion ounces. Your storage protocol that you select is outside the realm of what vsphere manages directly; they haven't needed to put a checkbox in there for iSCSI, NFS, FCoE etc since storage protocols are, in general, managed separately.
Ideally, you should configure your interface vmkernal on your Standard vSwitch or vDistrubuted switch with network details you have specified for the subnet / VLAN that you chose for NFS [skip the check boxes]. The final State, all your NFS traffic will then flow through the physical network links that are configured for your traffic NFS on your SvS/vDS.
Without knowing the details of your files and their abilities network/host/server configurations, Setup and best practices may vary.
-
Identification of customer's Windows 7 NFS UID GID information
I am trying to connect to the Windows 7 Client NFS on a server running on a computer (VxWorks) NFS. I am able to properly connect Client NFS software by a 3rd party on the NFS server. However, when I use the Client for NFS provided by Windows 7, I am unable to connect. The \\ip address\share Z: mount"command fails with the error code"the path not found network ". I can't do a ping of the computer running the NFS server.
The NFS Client operating system: Windows 7 Ultimate, 64-bit
Data captured by Wireshark
MOUNT V1 EXPORT call 3rd party client
Identification information Flavor: AUTH_UNIX (1)
Length: 32
Stamp: 0xc7065970Machine name: PC
UID: 1000
GID: 1000MOUNT V1 EXPORT appeal of the NFS client
Identification information Flavor: AUTH_NULL (0)
Length: 0It seems that the credentials of NFS Client are not correct. How can I change the flavor of AUTH_UNIX and the UID and GID to 1000?
Hello VDAEMP,
As Eddie and Sudarshan has said, the Microsoft Answers community focuses on issues and problems related to the consumer environment. Please join the public IT pro TechNet forums below:
TechNet - Windows Server
Thank you -
tare on server for NFS mangles of symbolic links
Hello
I'm running into a problem where symbolic links are distorted when calibration files that reside on the Server 2003 and Server 2008 running server for NFS, serving to Redhat Linux 4.6, 5.3, 5.7 and machines Ubuntu 11.10. Note that the test below, region of origin of the user is mounted on a server for NFS 2008 R2 installation. The question is not if I create a tar with files on the local computer. Absolute links are unaffected, related links. Help solve this for links are greatly appreciated.
[Dd ~/TEST]$ if = OneGig.txt of = / dev/zero bs = 1024 count = 1000]
[~/TEST]$ ln-s OneGig.txt test
[~/TEST]$ ln-s /home/USER/TEST/OneGig.txt test2[~/TEST]$ ls-l
Total 1000
-rw - r - r - 1 1024000 24 August at 14:11 OneGig.txt
lrwxrwxrwx 1 0 24 August to 14:12 test-> OneGig.txt
lrwxrwxrwx 1 0 24 August to 14:12 test2-> /home/USER/TEST/OneGig.txt[~/TEST]$ cd...
[~] $ tar cvf TEST.tar TEST.
TEST /.
TEST/OneGig.txt
TEST/test
TEST/test2
[~] $ tar tvf TEST.tar
drwxr-xr-x 0 2012-08-24 14:12:55 TEST.
-rw - r - r - 1024000 2012-08-24 14:11:29 TEST/OneGig.txt
lrwxrwxrwx 0 2012-08-24 14:12:46 TEST / test-> O
lrwxrwxrwx 0 2012-08-24 14:12:55 TEST / test2-> /home/USER/TEST/OneGig.txtThis forum is for users of the home computer having problems with files in the Windows consumer editions. Help server, check with the people in the TechNet forums, this is where the guys TI and server gurus hang out.
-
What "unmapped user access" for NFS share in Windows Server 2008 R2?
In the beginning, I followed this Youtube video to set up an NFS share in Windows Server 2008 R2:
https://www.YouTube.com/watch?v=BQ8Q_vsiksg
Please note that the default option to allow to access Unix unmapped user is not changed (to about 1 min 57 sec).
Then I found another article at http://www.vmwarearena.com/2012/07/create-nfs-datastore-for-esx-in-windows.html
But in this article, I like to select allow anonymous access (instead of allow access Unix unmapped user).
So who is correct? Or both are correct (which means that this 'unmapped user access' is not serious)?
And why should we have to allow root access (in the NFS share permissions dialog box)? What a security risk?
Thanks in advance.
Have you tried to run through the steps taken from the blog of VMware?
Also, there are 3 KB articles to know in case of potential problems
1. http://kb.vmware.com/kb/1004490
2. http://kb.vmware.com/kb/1003967
3. http://technet.microsoft.com/en-us/library/cc753302 (WS.10) .aspx
Noting in particular this section
Share the folder of Windows for NFS
To share the folder of Windows for NFS:
- Right click on the local folder you want to share via NFS.
- Click NFS share.
- Type the name for the share. For example, NFS-VMFS01.
- Delete to allow anonymous access.
- Click on permissions.
- Change the type of access in reading + writing and select allow root Access.
-
Hello
What is the maximum size for NFS and VMFS (iscsi and FCP) data created on vSphere 4.1, 5.0, and 4.0 stores?
Thank you
Tony
Hi Tony,.
You should find the answers in the various maxima of Configuration documents:
- http://www.VMware.com/PDF/vsphere5/R50/vSphere-50-configuration-maximums.PDF
- http://www.VMware.com/PDF/vSphere4/R41/vsp_41_config_max.PDF
- http://www.VMware.com/PDF/vSphere4/R40/vsp_40_config_max.PDF
Good luck.
Concerning
Franck
-
For NFS Datastore vSphere alarms
So I would like to create an alarm that corresponds to a (not state) event that fires when a NFS data store is disconnected. I found the trigger "Interruption of the connection to the NFS server", but it doesn't seem to work at all. Also, I would only triggers the action when the host is not in Maintenance Mode, because that would be very annoying for an outgoing call because a host has restarted for patches and generated alarm type disconnected "NFS Datastore.
use triggers esx.problem.storage.apd. *.
When NFS disconnects you will get official messages in the file vmkernel.log on the host
-
subnet VMkernel iSCSI for direct-attached storage
Hello
I have two hosts esxi 6.0 that I connect to a shared via iSCSI storage. I followed this document technical http://www.vmware.com/files/pdf/techpaper/vmware-multipathing-configuration-software-iSCSI-port-binding.pdf#sthash.7bztDoAS.dpf to configure interfaces iSCSI on the esxi with binding ports host. No problem.
I use HP MSA 2040 with a dual controller and I'll connect two hosts directly on the shared storage. As for the hardware, according to the documents on best practices, it is recommended to follow a vertical subneting for ports of controllers. However, I have not found no requirement for iSCSI ports in vmware, all of the examples I found configure each vmkernel port in a different subnet, as I did:
vSwitch0
iSCSI1 - vmk2 - 192.168.1.11 (unused vmk3)
iSCSI2 - vmk3 - 192.168.2.11 (unused vmk2)
So, my question is vmkernel ports are in different subnets? or could they be in the same subnet with no difference performance wise?
Any help would be appreciated.
Thank you!
in your case, because you are trying to achieve Multipathing to iSCSI Software initiator, the vmkernel ports should be in the same subnet.
-
I'm under NFS in a test lab. I have a unique storage system of NetApp 2240 running NFS. I don't have that one physical to work with, but like many switch switch ports as I need. The 2240 has two redundant controllers. Here are my thoughts:
-Use 4 wires total ethernet. Run two wires to controller 1 and t two cables to controller 2.
-On controllers - use e0a e0b on each contoller together in a virtual interface
-Use etherchannel between the storage system and the physical switch
Since I don't have that physical to work with only one switch is there a better way to do it?
Thank you!
Sounds like the best you can do with a single switch. I would opt for LACP instead of a static OFFSET, however.
-
I'm trying to backup on a NFS server using the script ghettoVCB.sh, but am not sure what I need to put the path. Here is the snippet of script:
directory that all VM backups should go (for example/vmfs/volumes/SAN_LUN1/mybackupdir)
VM_BACKUP_VOLUME=/vmfs/volumes/dlgCore-NFS-bigboi.VM-backups/WILLIAM_BACKUPS
NFS "ubuntu" server is 192.168.1.200 with a share called / files.
On the ESXi 4.0 Server, the default data store points to the local disk. Using vSphere client, I created a second named Ubuntu by NFS data store and robbed on the server (192.168.1.200:/files) successfully.
What am I supposed to use the path VM_BACKUP_VOLUME?
VM_BACKUP_VOLUME = / vmfs/volumes/Ubuntu /...
-
Create a new port of vmk on a vSwitch, add it to a group of specific ports and enable vMotion. If a group of ports corresponding to the specified name does not exist then it is created.
1New-VMHostNetworkAdapter -VMHost <host> -PortGroup <PG Name> -VirtualSwitch <vswitch> -IP <IP> -SubnetMask <Mask> -VMotionEnabled:$true -Confirm:$false
You can enable or disable a vmk for traffic management, vMotion or fault tolerance. All are disabled by default and each of them can be activated when the vmk is created, similar to how vMotion has been enabled in the above example. To disable just change $true to $false.
12
3
Get-VMHostNetworkAdapter -VMHost <host> -name <vmk> | Set-VMHostNetworkAdapter -ManagementTrafficEnabled:$true -confirm:$false
Get-VMHostNetworkAdapter -VMHost <host> -name <vmk> | Set-VMHostNetworkAdapter -FaultToleranceLoggingEnabled:$true -confirm:$false
Get-VMHostNetworkAdapter -VMHost <host> -name <vmk> | Set-VMHostNetworkAdapter -VMotionEnabled:$true -confirm:$false
Change the IP address of a vmk
1Get-VMHostNetworkAdapter -VMHost <host> -name <vmk> | Set-VMHostNetworkAdapter -IP <ip> -confirm:$false
And delete a vmk
1
Get-VMHostNetworkAdapter -VMHost <host> -name <vmk> | Remove-VMHostNetworkAdapter -confirm:$false
===
I am creating above
Nowhere I'm option specify the vlan... IE no mgmt ft/no vmtion/no on the vmk also in my case, that nothing will be
Please suggest
When you have the portgroupname, you can use the Set-VirtualPortgroup cmdlet to create/modify the VlanID.
-
Why vmkernel port for iscsi and vmotion services?
Hello
Just a quick question.
I was trying to understand - why we need to have a vmkernel port when we want to use features such as vmotion or iscsi.
Why not use them with a console port service or a port of vm management traffic.
No particular reason.
Thank you
Yes. You're not really all that traffic produced by DRS and vMotion actions in your production (maybe time critical).
Concerning
-
ESX/ESXi 4.1 support for NFS v4
Hi team
Just out of curiosity, ESX/ESXi 4.1 already does support NFS v4? Or it only does support NFS v3?Nevertheless I've invited the VACP rests always on ESX 4.0, but correct me if I'm wrong.Thanks in advance.Jose Maria GonzalezAuthor of 101 secrets of VMware vSphereVCP4 VCP3, VCP2, RHCE, VCI, MCSE, vExpert 2009, 2010I don't know that NFS v3 over TCP is supported.
André
Maybe you are looking for
-
Charges Post Huff in Firefox but will not load stop and cannot be read. Not the same with IE
Huffington Post is the site that is unable to stop loading. (there may be others, but I don't know them)I tried everything that was suggested and without success. I read every day Post Huff and must use Internet Explorer to be able to read and click
-
How to connect to the internet via the wireless?
I just bought my first laptop Toshiba L10, and I'm unable to start, how do I connect to the internet using the wireless feature and no track? More frustrating is how can I connect to the internet in the first place, wireless.Help, please.
-
Extension memory for Tecra M5 PTM50E-07H02M1D
Hello Can someone tell me if memory 2 GB PC2 DDR2 (667 MHz), PA3513S-1M2G, is suitable for Tecra M5, Model No.. PTM50E-07H02M1D? Thanks in advance
-
Hello I'm trying to set up my new 4500 want on my windows 8 PC, wireless. I connected the printer through the wireless on the printer setup screen to the network. I think it's successful because the print report said "without problems" and I have a l
-
Update from 0.9.3 to 0.9.4 (Sim &; SDK)
Hi all! What is the recommended way to upgrade 0.9.3 to 0.9.4? If 0.9.4 can be installed in place that upgrades the files or do I have to uninstall in Control Panel? I have already created my signature keys - what files should I keep/copy/prevent get