Behavior of UCSM vNIC/vHBA placement policy

hi,
Can anyone please explain me the behavior of UCSM vNIC/vHBA placement policy, and its three options assigned-only, Exclude-Dynamic and Exclude-Unassigned. I have gone through the documentation, but I am not very clear yet.
May be an explaining with an example will help.
Sample scenario.
Full blade having Palo on left side, and Menlo Emulex on right side, with 8 VNICs (4 dynamic, and 4 manual), with each of the three options.
thanks
Tejo

/* Style Definitions */
table.MsoNormalTable
{mso-style-name:"Table Normal";
mso-tstyle-rowband-size:0;
mso-tstyle-colband-size:0;
mso-style-noshow:yes;
mso-style-priority:99;
mso-style-parent:"";
mso-padding-alt:0in 5.4pt 0in 5.4pt;
mso-para-margin-top:0in;
mso-para-margin-right:0in;
mso-para-margin-bottom:10.0pt;
mso-para-margin-left:0in;
line-height:115%;
mso-pagination:widow-orphan;
font-size:11.0pt;
font-family:"Calibri","sans-serif";
mso-ascii-font-family:Calibri;
mso-ascii-theme-font:minor-latin;
mso-hansi-font-family:Calibri;
mso-hansi-theme-font:minor-latin;
mso-bidi-font-family:"Times New Roman";
mso-bidi-theme-font:minor-bidi;}
Tejo, there are two reasons that you would want to use a placement policy.
Full width blades with 2 mezzanine cards - When you have 2 mezzanine cards a placement policy allows you to specify which mezzanine card to assign a vNIC/vHBA to. By doing this you can better load balance the network and storage traffic. In a placement policy the 2 mezzanine cards are identified as vCon1 and vCon2.
Dynamic vNIC Connection Policy – When you have a Dynamic vNIC Connection Policy as part of a Service Profile you want to have a placement policy so that the dynamic vNICs are created below the standard vNIC/vHBAs in the PCI order. Dynamic vNIC connection policies are used in the configuration of VN-Link in hardware. This type of policy requires the Palo mezzanine card though.
Other than those two reasons I don't know when you would use a placement policy unless you just wanted to make sure the OS sees the vNICs/vHBAs on specific PCI slots.
The exclude-dynamic comes into play with the Dyamic vNIC Connection Policies I mentioned in option 2 above. Exclude-unassigned would be where you have some that you want to assign (vNICs) but you don not care about where the vHBAs get assigned.
I personally have only had the need to use them in the 2 scenarios above.
With your specific scenario you would use the placement policy to assign some vNICs/vHBAs to the Emulex card (2 max for vNIC and vHBAs) and some to the Palo (dynamic vNIC connection policy or more than 2 vNICs. Or you might want all of your vHBAs to use the Emulex and all of your vNICs to use the Palo.

Similar Messages

C460 Rack Server with two VIC 1225 - UCSM vNIC placement policy

I am trying to integrate C460 having 2 VIC 1225 cards with UCSM. The objective is to have 2 ethernet interfaces assigned to each VIC , one per FI.
vNIC 1 on VIC 1 - FI A failover to FI B
vNIC 2 on VIC 1 - FI B failover to FI A
vNIC 3 on VIC 2 - FI A failover to FI B
vNIC 4 on VIC 2 - FI B failover to FI A
I have created a service profile with 4 vNICs and assigned to C460. When the UCSM is configured to use the system default for vNIC/vHBA placement policy, all 4 vNICs are assigned to one VIC. When trying to place them manually with 2 on vCon1 and 2 on vCon2, I am seeing insufficient resource error. Both the VIC are recognized and UCSM server tab shows both of them as adapters.
Could someone please let me know how to achieve it?
Thanks!

The C460 has an integrated BCM 57711 which would also be recognized by the system. What's most likely happening is that the VIC 1225s are on vCon1 and vCon3 and the BCM card is on vCon2, since FF is not supported on the Broadcom adapter this would cause the association to fail. Try placing the manual placement with vCon1 and vCon3 and see if this clears up the issue.
Steve McQuerry
UCS - Technical Marketing

VNIC Placement Policy for VMWare

I'm setting up a new UCS system to run ESXi 5.0 and would like to set up the service profile so that the vNICs get placed in a particular order. I've looked at the vNIC/vHBA placement policy and I think that between this and the service profile this is where I want to configure the required placement. What I'm not sure of is how to configure the placement policy for optimal efficiency.
I have a VIC 1280 card in each B230 M2 blade. I have 12 vNICs (6 on Fabric A and 6 on Fabric B) that I'm creating for the various networks (Prod, DMZ, Management, vMotion, etc...) on each server. When I look at my current vNIC placement on one of the ESXi hosts where I let the system place the vNICs it looks like they are all on vCon1.
Shouldn't my vNICs be split across the vCons? If so how should I split them? Any best practices or suggestions on how I should configure my placement?
Thanks.

This question has been answered pretty heavily here on CSC and on other blogs - so I wont go into great detail.
In short:
vCon's refer to Adapters. One adapter (1280) = 1 vCon.
PCI ordering is managed within the vCon by vNIC/vHBA placement Policy.
Regards,
Robert

Adding vNICs to vNIC LAN Connectivity Policy via UCS Central PowerTool (BETA)

When I try to add an vNIC to an VNIC LAN Connectivity policy by using the UCS Central Powertool (v0.94) I getting the following error:
D:\Scripts> Add-UcsCentralVnic -Name vmnic0 -VnicLanConnPolicy test
Add-UcsCentralVnic : ucc01:Policy org-root/lan-conn-pol-test is resolved from remote policy server. Create/Delete/Modify operations are not allowed.
At line:1 char:1
+ Add-UcsCentralVnic -Name vmnic0 -VnicLanConnPolicy test
+ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ CategoryInfo : NotSpecified: (unidentified-fail:String) [Add-UcsCentralVnic], Exception
+ FullyQualifiedErrorId : 170,Cisco.UcsCentral.Cmdlets.AddUcsCentralVnic
whatever I try, I still get the same error every time ("Create/Delete/Modify operations are not allowed").
I'm using UCS Central version 1.2(1a) and UCS Central PowerTool version 0.94 (BETA)

Maybe you post your problem in https://communities.cisco.com/docs/DOC-53127 where other issues related to UCS Center powershell are discussed.

Install all drivers + vNIC/vHBA configuration on W2k8R2 on B200M3 server

Hi everybody,
I am a network engineer and rather a newbe on system…
I have just installed a w2k8R2 64bit on a B200M3 in our new UCS architecture (local storage; W2k8 as bare metal OS)
3 questions:
What is the best way to install all W2k8R2 drivers for this B200M3 server ? At present, I have many unknown hardware; I have just install manually the drivers for VIC 1240 by using ucs-bxxx-drivers.2.0.3b.iso or/and ucs-bxxx-utils-windows.2.0.3.iso. I looked for a exe to apply all W2k8R2 drivers for B200M3 but I could not find it.
The Service Profile of this server allowed 3 vlans : 1 natif + 2 tagged. In the properties of the network card under W2k8, I could not have any option to active the tagged. How can we do that?
The Service Profile of this server has 2 vHBA. How can i configurate the failover between fabricA and FabricB
Many thanks in advance.
Nicolas.

Hello Nicolas,
You are on the right track. Mount the ucs-bxxx-drivers.2.0.3b.iso image via vmedia and browse it from Windows OS.
1) Intel Chipset
2) FNIC ( HBA )
3) ENIC ( Network )
Currently, there is no one-click option to install necessary drivers based on server model. The following link helps you to navigate the ISO folder
http://www.cisco.com/en/US/docs/unified_computing/ucs/os-install-guides/windows/BSERIES-WINDOWS_appendix_0100.html
If you have many such installations to do, then you might want to create custom W2K8 image with necessary drivers built into it.
2) VLAN tagging
I assume you have single vNIC that allows 3 VLANs and want the OS to tag the traffic. I do not see VLAN tagging in ENIC driver properties. I will find more information on it.
If not, you can create three nics and UCS vNIC will tag the traffic as defined in the service profile.
3) HBA failover is handled by Multipath failover driver in OS( MPIO ) . Depending upon the storage array, you can install appropriate MPIO driver.
HTH
Padma

Problem with vhba and dynamic vnic placement

Hello all,
I have a problem with vnic/vhba placement policies on an UCS-B:
When a configure a Service Profile, evrything is good:
- I use a "specify manualy" vnic/vhba placement policy and can create this configuration:
- vhba 0 : Order 1
- vhba 1 : Order 2
- veth0 : Order 3
- veth1 : Order 4
- veth 2: Order 5
- veth 3: Order 6
- 8 Dynamic vnic : Order 7 to 14
As soon as i associate this SP with a B200M2 (with palo), UCS manager change this order like that:
- veth0 : Order 1
- veth1 : Order 2
- veth 2: Order 3
- veth 3: Order 4
- 8 Dynamic vnic : Order 5 to 12
- vhba 0 : Order 13
- vhba 1 : Order 14
I've tried several placement policies (auto, exclude dynamique from vcon 1, ....) but it's still the same.
I need to maintain previous order with first vhba, then vnic and finaly dynamic vnic...
My config:
- 1* UCS 5108
- 4 *UCS B200M2 with palo adapter
- UCS MANAGER 2.0(1q)
Does someone know this problem and how to solve it ?
Thanks in advance
Regards
Sebastien.

Sebastien,
When using dynamic vNICs you can't define more than one physical NIC (eth0 and eth1). If you try to create more than two vNICs (using a single VIC) with a Dynamic vNIC policy, it should error out on you during association.
The vCon connection policy is really only used for full width blades with dual mezz cards populated. This gives you some control as to which mezz card and interface vNIC/vHBAs are placed on.
With a B200, the only virtual PCI slots you'll have will be 1 and 2. To my understanding you need the vHBA's as the lowest PCI device # - is that correct?
For your placement policy, set slot 1 to be "Assigned Only", and leave the rest set as "All". Next assign your vHBAs both to Slot 1, and click ok. Associate the Service Profile to the blade and report the results.
This should set the desired placement and ordering for your vHBAs as the lowest.
Regards,
Robert

VCon Placement in UCS

Hi there,
I have a question in regard to placement of vnic's in UCS specifically to the vCon assignment.
In my system I have B200M3 servers with the UCS Vic1240+mlom port expander card.
What I'm wondering is how many vCon's I have in my system? From what I have read, each vCon is the virtual representation of the physical adapter, so in my case I would have a single vCon. This would make sense, because when I look at where the system has placed my vnics, they are all associated to vCon1.
However, when I look at the "Modify vNIC/vHBA Placement" and do a specify manually, I see four vCons (vCon1 - vCon4). Are these there by default, and/or should I be able to use vCons 2-4? If I am able to, then I wonder why the system has only chosen to use vCon1 ?
Any assistance in this understanding would be gratefuly received!
Thanks
Dominic

The only question remains is; Suppose you always want to have a specific order with one adapter, but once you move a Service Profile to another server with two adapters, you want them to be spread amongst these two adapters in a round-robin fashion. For example:
One Adapter: vCon 1:
1 - iSCSI_Eth0
2 - iSCSI_Eth1
3 - Eth0
4 - Eth1
5 - VM_vEth0
6 - VM_vEth1
Now if you move above configuration to another server with two adapters:
First Adapter: vCon 1:
1 - iSCSI_Eth0
2 - Eth0
3 - VM_vEth0
Second Adapter: vCon 2:
1 - iSCSI_Eth1
2 - Eth1
3 - VM_vEth1
Is this correct? I think I have to alway spread them amongst two vCons or select automatic placement, right? With boot from iSCSI this can be tricky.
Boudewijn

ESX uplinks on a UCS Blade

After installing ESX on UCS blade ( adapter type is Palo, number of nics is 2) sometimes I see that the uplinks are vmnic2 and vmnic3 (output of esxcfg-nics -l)
Where are the output of esxcfg-vswitch -l shows uplink as vmnic0 and it also throws a warning saying vmnic0 does not exist.
This does not happen all the time. Anyone know why this could be happening?

I've seen this happen quite a bit when you don't have the nics as the first devices in the vNIC/vHBA Placement. By default we put the HBAs first sometimes ESX doesn't like this and will put the NICS as vmnic2 and 3. If you move the nics above the HBAs they come in as vmnic0 and vmnic1. I always reorder and put my nics before my storage.
louis

VNIC templace change wants to reboot blade

I have a vNIC template for VMware. The vNIC template has a bunch of VLANs that map to port groups on ESXi servers. I have those vNIC templates in my service profile template that I have used to create some service profiles for ESXi servers. I unbound the service profile template from the service profiles right after creation. When I want to add a port group to an ESXi server I got to UCS and add a VLAN then add that VLAN to the vNIC template that is associated with the service profile template and service profiles of the ESXi servers. That has been working successfully for a year. Recently, I cloned the service profile template because I needed to create a new profile for some ESXi servers that I'm building. I had to make changes to the vHBA templates in the service profile template. I didn't make any changes to networking configuration in the cloned service profile template. This morning when I added a VLAN in UCS and went to update the vNIC template with the new VLAN it prompted for a reboot of the ESXi servers that were created from the cloned service profile template, but not the other ESXi servers that I created with the original service profile template. I've compared the service profile templates but can't see any difference (other then what I purposefully changed in the vHBA templates).
What could I be doing wrong that would prompt for the ESXi servers to reboot? I should be able to add vlans to the vNIC templates without a need for a reboot of the host.

Hello Russ,
Are you using initial or updating template for vNICs /s ervice profile ?
I did a quick test in my lab and observed the following behavior.
UCSM 2.0.1t
1) Created Initial vNIC template with certain VLANs
2) Binded the vNIC template to an existing service profile that is associated with a blade.
3) Unbinded the vNIC template
4) Added VLANs to vnic template
5) Binded the nics to vNIC template and NIC is updated with new list of VLANs
Above steps did not reboot the server.
6) Cloned the service profile and associated it with a blade.
7) When trying to add vnic template to the service profile, it prompted for reboot
8) After reboot, steps 3 to 5 did prompt for reboot.
Does it ask for reboot every time you add a VLAN to vNIC template or just during the initial configuration as observed above ?
Padma

VNic Mac addresses not associated correctly when installing OVM Server

We are trying to get our first Oracle VM environment running and noticed that on the UCS the mac assocated with eth0 ended with 0B but on the Oracle install the eth0 interface mac address ends with 08. The mac for eth3 on the Oracle install lines up with what the UCS sees as eth0. How do we correct this and ensure the interfaces line up so that eth0 on the UCS is eth0 on the Oracle service profile ?

This is going to depend on how Oralce numerates the PCI BUS. That will dictate which PCI device the OS sees in which order.
You can try manipulating the order using a vNIC placement policy. This will allow you to ensure certain vNIC/vHBAs appear first on the PCI BUS for the host.
Regards,
Robert

VNIC Manager arbitrary MAC addresses?

Hi,
is there a way to enter arbitrary MAC addresses using the VNIC Manager in OVM Manager 3.0.3? I need to migrate a few machines to OVM (both physical and virtual) where I need to maintain the existing MAC address. Unfortunately the WUI only allows to specify the last 3 words of the MAC. Is there any way around this so I can specify the complete MAC?
Thanks,
Bernd

This is going to depend on how Oralce numerates the PCI BUS. That will dictate which PCI device the OS sees in which order.
You can try manipulating the order using a vNIC placement policy. This will allow you to ensure certain vNIC/vHBAs appear first on the PCI BUS for the host.
Regards,
Robert

[Solved] Openbox Window Placement not in Center

In obconf, under Windows, I have:
√ Focus new windows when they appear
√ Center new windows when they are placed
Prefer to place new windows on: The monitor with the mouse
Primary monitor: Monitor With Mouse Pointer
But in the corresponding rc.xml, I have
<placement>
<policy>Smart</policy>

<center>yes</center>

<monitor>Mouse</monitor>

<primaryMonitor>Mouse</primaryMonitor>

</placement>
And the behaviour is this:
Windows try to get as close to the top left corner as possible without overlapping any other windows. This does sound like what "Smart" policy would refer to, but it is not listening to the <center>yes</center> tag.
Last edited by jameh0 (2014-03-22 17:59:40)

drcouzelis wrote:
drcouzelis wrote:it's not hard to patch, so no problem.
I just finished patching Openbox to restore the former random window placement behavior. You can find it here: openbox-git-random-placement
It works but I didn't fully test all scenarios (multi-monitors, different window states, and so on...) so if you find any bugs in my patch or random window placement code please let me know!
Im having trouble discerning what exactly you mean. I could build your version to find out, but the current openbox-git seems to work exactly this way?
I have smart placement and center set. When I open a new window on a blank workspace, its placed centered. If I open another, it will be placed in empty space. It will keep placing in empty space until the desktop is completely hidden, and then it will start randomly placing the window.
I have considered talking to the Openbox devs about possibly implementing the placement algorithm of Fluxbox- Flux will evenly distribute the windows even after the entire screen is filled/covered.
**EDIT** Do you mean your version starts to random place the moment it cant fit the window in empty space without even partially obscuring an existing window?
Last edited by GSF1200S (2014-03-27 04:18:56)

Jax-ws: dynamic proxy-suppress weblogic's default ws-policy handling

hi all,
i am writing my own jax-ws client (dynamic proxy) to call a web service, secured by usernametoken profile.
the wsdl of the web service do include the ws-policy information.
sample client side code:
URL url = new URL("path_to_wsdl_file");
QName serviceQname = new QName("http://model/", "CalculationService");
Service service = Service.create(url, serviceQname);
//add resolver and jax-ws handler to add username token SOAPElement fragment to the SOAP message.
the above code runs fine in J2SE 6.0 environment (without any weblogic-specific library).
but when i run the above code in a servlet (running in weblogic), the above code fails with the following error:
Exception in thread "main" javax.xml.ws.soap.SOAPFaultException: Unable to add security token for identity, token uri =http://docs.oasis-open.org/wss/2004/01/oasis-200401-wss-username-token-profile-1.0#UsernameToken
at com.sun.xml.ws.fault.SOAP11Fault.getProtocolException(SOAP11Fault.java:196)
at com.sun.xml.ws.fault.SOAPFaultBuilder.createException(SOAPFaultBuilder.java:122)
at com.sun.xml.ws.client.sei.SyncMethodHandler.invoke(SyncMethodHandler.java:119)
at com.sun.xml.ws.client.sei.SyncMethodHandler.invoke(SyncMethodHandler.java:89)
at com.sun.xml.ws.client.sei.SEIStub.invoke(SEIStub.java:118)
at $Proxy23.add(Unknown Source)
at client.ServiceLocator.main(ServiceLocator.java:32)
it seems that the weblogic server will try to handle the ws-policy element (according to those in the wsdl), even though
i have already added the username-token security header by my own code.
As i don't add some weblogic-specific code (e.g. weblogic.wsee.security.unt.ClientUNTCredentialProvider), the web service call fails.
is it possible to disable/suppress such behavior (weblogic runtime handle the ws-policy element) per web service call???
(i don't want a server scope solution, as there may be other web service client running on the same weblogic server that rely on such behavior).
i don't want to use weblogic's API (e.g. ClientUNTCredentialProvider) as i want my web service client be portable (username token is easy to be handled by my own code)
thank you.

Hi René ,
i just use plain HTTP, not HTTPS.
seems don't involve any certificate, am i right?
in J2SE environment, i can run the webservice client without problem.
but once i run the client in weblogic, the problem mentioned occurred.
if i manually delete the ws-policy related tag in the wsdl file (i download it to local filesystem for testing), the client can run in weblogic too.
so, think there is some setting in weblogic that try to handle the ws-policy by its so-called 'credential provider'(e.g. ClientUNTCredentialProvider ?)
thank you.

More ZPM in ZCM questions--intended behavior?

Perhaps I am doing something wrong or not looking in the correct spot.
If I use ZPM in ZCM, there seems to be some deficiencies:
a) Refresh of patch status. Meaning I go into Patch Management, click on a patch and select Actions -> Update Cache. I get the icon to change animation but it NEVER changes status unless you manually refresh the web browser. I do not believe that ZPM 6.4.x (Novell branded product) behaves this way. I think the ZPM 6.4.x will refresh the page automatically for you so you know when it's done downloading.
b) Patch deployment status - you can't really tell what's going on with the devices. In ZPM 6.4.x I can go into devices and I see animations as to what the machine status is (is it on, is it doing a DAU, is it deploying a patch). I don't seem to be able to find any of this information in ZCM (meaning I can't just look at the device icon and see what it's doing, I have to go into the patches tab or something to see if it got deployed, but I don't know what the "live" status of the machine is).
c) Patch Deploys in general - I am pretty sure I can go into ZPM 6.4.x and find my deployment tab and see what I chose to deploy, when, and to which machines. I don't seem to be able to find this in ZCM. I do see something like "deploy on some heinous formatted date/time" but that's it. I have to scrounge around inside the deploy to the actions section and even then it doesn't reall show me in a nice format which machines it was deployed to, let alone the status of the machines.
Again, I could be looking in the wrong spot, or perhaps it's not there.

In response here we've gone through great efforts to make the integration consistent with ZCM behavior.
a) Only a couple of pages do auto refresh of status in ZCM. This is to allow for the greatest performance. ZPM is consistent with the rest of the ZCM behavior such as bundle status and policy status. By refreshing the page status updates.
b) Since all deployments are done using the bundle system in ZCM you can track all status of ZPM deployments by looking at the Bundle status. There is a ZPM Assignment bundle created for each deployment. It should have status of deliveries and errors if you look at it.
c) You should be able to go to the ZPM Assignment bundle and look at the Install Action Set. It will show you which patches are being deployed.

Initial window placement openbox

Hi all,
Whatever setting I'm using in openbox, my new windows are placed at the top-left corner of the screen. Hoewever, I have a dzen2-conky monitor-bar at the top of the screen with a height of approx 20px. Unfortunately, my new windows are placed with their title bar under the monitor-bar. I can move those windows only by pressing the ALT key and moving the mouse.
<placement>
<policy>Smart</policy>
<center>yes</center>
</placement>
How can I fix this so openbox is:
1. placing my new windows in the middle of the screen
or
2. placing the windows wherever it wants them, but not in the topmost 25 pixels of the screen
Last edited by Ibex (2009-10-28 15:07:35)

Sweet, that's what I was searching for in the #2 solution. Thanks
Edit; damn, this seems to work fine for maximized windows, but it does not work for new smaller windows like a terminal (Eterm in my case). This window is still placed in the top 20px of the screen .
Edit2; seems like Eterm is the only program having this issue anyway...
Edit3; damnit, it seems to be the Eterm theme that is forcing the window to spawn at 10px from the left and 10px from the top. Editting the Eterm theme was the solution.
Last edited by Ibex (2009-10-28 15:43:59)

Behavior of UCSM vNIC/vHBA placement policy

Similar Messages

Maybe you are looking for