cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
5151
Views
5
Helpful
12
Replies

B200M4 w/1340 vNIC placement order not adhered to when you go from 6 vNICs to 8 or more vNICs

zepherdog
Level 1
Level 1

Hi,

I'm running into an issue when I increase the vNIC qty on my blade servers from 6 to 8 or more vNICs and the placement order is not adhered to.

Basically the first 3 vNICs and the last are in order, from 4 onwards its mixed up........

Right now it stops the NFS datastores from mounting as the NFS nics are associated to a different vSwitch and are some of my other NICs

I have tried the following placement order methods - all with the same results

  • Let System Preform Placement
  • Specify Manually
    • with and without changing the Admin Host Port 
  • Placement Policy
    • RR 
    • Linear Order

 

I'm running an A and B MAC pool (Assignment order = Default) and my A and B vNICs were associated accordingly

I've created a new larger MAC pool and set the Assignment Order to Sequential, changed all the vNIC tempaltes to use the new MAC pool, created a new SP-Template and new SP, asscioated the server - still no success

Ive upgraded firmware from 2.2(3g) to 2.2(5b) - no success

 

 

Any ideas?

Cheers

Mike

 

1 Accepted Solution

Accepted Solutions

Saurabh Kothari
Cisco Employee
Cisco Employee

Hello,

CSCut78943

https://tools.cisco.com/bugsearch/bug/CSCut78943/?reffering_site=dumpcr

follow the workaround as per bug:

CSCuv34051

https://tools.cisco.com/bugsearch/bug/CSCuv34051/?reffering_site=dumpcr

Kindly let me know if theres' any query.

Regards,

-SK

View solution in original post

12 Replies 12

Saurabh Kothari
Cisco Employee
Cisco Employee

Hello,

CSCut78943

https://tools.cisco.com/bugsearch/bug/CSCut78943/?reffering_site=dumpcr

follow the workaround as per bug:

CSCuv34051

https://tools.cisco.com/bugsearch/bug/CSCuv34051/?reffering_site=dumpcr

Kindly let me know if theres' any query.

Regards,

-SK

I posted a topic in this forum almost a month ago about this. 

 

I also just closed a TAC case about this. There isn't much Cisco can do about this as its caused by how ESXi deals with enumeration. 

 

It basically came down to my work around (set all VNICs and VHBAs to admin host port one on vcon one) or having to delete all the adapters and reading in esxi anytime you add VNICs. If you go with my work around you don't get the benefit of multiple pci connections. If you go with the VMware method you have a significantly manual process that must be done on every update of nics. 

 

Its not ideal and I wish VMware would handle it better. Hopefully they introduce something in a future release. 

Hi Steven, thanks for the update - would you mind if i got the TAC case # so i can reference this for my case?

Cheers

Mike

SR 635778227

 

The engineer provided me bug ID CSCuv34051

Thanks - that bug ID isn't even for our relevant hardware as noted above

When I apply the settings as you have mentioned in your post I'm still getting inconsistant  placement

Thanks for the reply 

Re CSCut78943

There is another bug that stops the AdminHost port due to other Policies - I created a new SP-Template with manual vNIC creation and specified the order - set all vNICs to vCON1, tried;

  • Any Admin Host port - Any
  • Admin host port - 1 

No success

 

 

Re CSCuv34051

There seem to be to many difference

B200M3 vs M4

2.1(3b) vs 2.2(3b) and 2.2(5b)

VIC 1240 vs 1340

 

Hey,

I totally agree with you on the bug.

However my intent was to provide the solution.

If you read the workaround it kinda solves tour problem.

It also mentions a VMware KB article.

Thanks,

-SK

I've looked into the VMware KB a bit more - the remove, boot, add, boot work around seems to work

Thanks

hey thank you.

can you mark the post as answered, it will help other people to find it easier.

thank you again.

Done - I've also documented my process in more detail to help

Cheers 

Mike

After reviewing the VMware KB this "workaround" worked for me

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2019871

  1. Review the esx.conf
  2. Create a new LAN Connection Policy with only one VNIC for vKernel Management
  3. Update the Service Template
  4. Reboot the host to accept the new VNIC configuration
  5. Review the esx.conf file to confirm that there is only one vmnic present and all others are delete
  6. Modify your Service Template in accordance with your new VNIC qty and desired layout
  7. Reboot the host to accept the changes
  8. Compare the vmNIC order to the UCS VNIC order - mine now matched

Cheers

Mike

 

abhishek643
Level 1
Level 1

We were running in to same issue, after adding new nics in UCS VMWare will go crazy. All networking will be in mess.

 

Used below link of VMware to reorder nics and it worked flawlessly after that

 

https://kb.vmware.com/s/article/2091560?other.KM_Utility.getArticleLanguage=1&r=2&other.KM_Utility.getArticleData=1&other.KM_Utility.getArticle=1&ui-comm-runtime-components-aura-components-siteforce-qb.Quarterback.validateRoute=1&other.KM_Utility.get...

 

Regards,

Abhi 

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: