Mark Minasi's Tech Forum
Sign up Calendar Latest Topics
 
 
 


Reply
  Author   Comment  
anthony

New Friend (or an Old Friend who Built a New Account)
Registered:
Posts: 56
Reply with quote  #1 
So this service has always irritated me as being half-ass. It's incomplete. If you are going to make your OS smart enough to determine which network it's on, why would you not allow me to "SAVE" NIC profiles and switch between them easily without re-entering IP info manually each time. But I digress...

We have an RDS Farm of 5 session hosts that run on our production Hyper-V cluster. It's encompasses a 2 cabinet Cisco UCS environment with 4 blades in each enclosure. It's an 8 node cluster. Obviously there is more running on this than the RDS hosts but just so you have the info I guess.

So very intermittently, the Network Location Awareness service decides that it want's to set the NIC to "Identifying..." and drops all the connections on the server. This can happen once a day. Once an hour. Once every 5 minutes... Obviously this is a bad thing as all the remote users get booted whenever this happens. When it gets to this point, the only thing that let's it "find" itself back to the domain network and restore connectivity is either a reboot of the server or disconnecting the NIC in Hyper-V and re-connecting it.

I found some posts like this one:
http://serverfault.com/questions/10014/disable-windows-server-network-locations

But, I was wondering if there was a way to just either disable this service, or force it to always be on our Domain Network? It seems that in the example above - setting it to be "Private" by default is really not helping.

__________________
If Chewbacca lives on Endor - You must acquit!
0
dennis-360ict

New Friend (or an Old Friend who Built a New Account)
Registered:
Posts: 94
Reply with quote  #2 
Ive had the same kind of problems, with isolated hosts (dell r630) and on win12r2. I found that the nic teaming (or the switches) are causing delays (probably because of timeouts) in starting nla. I found out that restarting nla after booting results in the correct domain profile. In my case setting the ip to dynamic instead of static helped.

Ive tested somewhat with setting services on delay start, but because of the dependencies ( i tried to delay them too), but changing these delay settings gave me intermittend results, so i would recommend them.

As far as i know (but im recollecting here) you cant disable nla or force the domain profile. You could however change the firewall rules for the other profiles like private/public, because i would guess the servers are always connected to the domain. But that would be just a workarkund instead of a proper solution.

However, im guessing this is a result of nic teaming and switch config. Weve seen the issue on unconfigured dell 6248, 2724 and x1052 whic are connected to other switches. The host have the independent vswitch config and hyperv port configured, some on dynamic. Im guessing stp/fstp is doing sonething to cause the delays. Im no network admin, but we are going to research this more with our new switches (dell n2024) for which ive hired a networking admin party. Im hoping to let them figure out if there are any delays or such on the switchport of the hyperv server. This will be on the 4th of march, so ill try to add to this post if i learn anything.

__________________
-----
Home is where is sleep
360ict.nl/blog
thegood.cloud
0
KirkLashbrook

Still Checking the Forum Out
Registered:
Posts: 7
Reply with quote  #3 
You probably want to troubleshoot how/why your NICs are dropping from the network and re-identifying the connection. That's not normal.

Once set, the network type is static and does not change.

You can turn-off NLA but that's a workaround, not the root cause.

0
anthony

New Friend (or an Old Friend who Built a New Account)
Registered:
Posts: 56
Reply with quote  #4 
I completely agree. I was just venting I guess. We have a ticket open with Cisco and Microsoft about this so hopefully we will find out something soon.
__________________
If Chewbacca lives on Endor - You must acquit!
0
wobble_wobble

Avatar / Picture

Associate Troublemaker Apprentice
Registered:
Posts: 940
Reply with quote  #5 
We had some wild crazy connectivity issues with SQL on Server 2012.
We disabled TCP Chimney, fixed the biggest issues/ dropping off the network and kicking in failover.

https://technet.microsoft.com/en-us/library/gg162682(v=ws.10).aspx

May be nothing, or may be worth a punt

More info 
http://blogs.technet.com/b/onthewire/archive/2014/01/21/tcp-offloading-chimney-amp-rss-what-is-it-and-should-i-disable-it.aspx

__________________
Have you tried turning it off and walking away? The next person can fix it!

New to the forum? Read this
0
dennis-360ict

New Friend (or an Old Friend who Built a New Account)
Registered:
Posts: 94
Reply with quote  #6 
that reminds me, there have been lots of issues with VMQ enabled, so turn that off too. It should only be enabled on 10Gb nics.

i wrote down how to do this here;
http://newforum.minasi.com/post/hyperv-guests-lose-network-connectivity-7928912?highlight=vmq&pid=1290919975

__________________
-----
Home is where is sleep
360ict.nl/blog
thegood.cloud
0
Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.