cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
Announcements
774
Views
0
Helpful
12
Replies
Highlighted
Beginner

Cluster over Wan sub admin page loads very slow

I have CUCM Publisher and Subscriber resides at our corporate office.  CUCMs are running on 8.5.1.15900-4 code.

I recently installed 2nd Subscriber at our remote office. Those 2 sites are connected via DS3 MPLS network.  Bandwidth is not an issue.

Some reason, logging to the remote site Subscriber server administration pages take forever to load.  I can access the Subscriber server at our Corp office works just fine.  Is there a bug on the code i am on?

1 ACCEPTED SOLUTION

Accepted Solutions
Highlighted

A.gooding makes a good point,

Other things that can cause slowness:

  • 1/2 duplex connections
  • Not disabling LRO on VMWare
  • Circuit oversubscription

Have you checked any of these?

Please help us make the communities better. Rate helpful posts!

View solution in original post

12 REPLIES 12
Highlighted
Enthusiast

According to the SRND the latency between any two nodes must not exceed 80ms round trip.  Since you have a DS3 it sounds like you have bandwidth. The question is do you have the appropriate round trip latency as well?

If you login to the console of the publisher, run the following command:  utils dbreplication runtimestate

The round-trip ping times will be listed. Ensure it is < 80ms. 

If that is fine, I'd next look for packet loss and or packet fragmentation issues.

Please help us make the communities better. Rate helpful posts!
Highlighted

The round-trip ping times is around 55-60 msec so i think i am OK.

i terminal in to the remote office server and have tried to access the server but same result.

i am running CUCM on VMWare at the remote office and Pub and Sub server on MCS server at Corp office.

Highlighted

Hi there

Is this on an mcs or ucs system. If ucs what's the vmware version. If mcs, I did have a somewhat similar experience a few years ago and after some checks my provider had specific port configs that were blocking Mac addresses which caused slowness initially and then of course db failure eventually.

Just thought I'd chime in

Ps. I also had an issue with the ucs hence the reason but this was an issue ith Irq mapping.

Thanks

Highlighted

A.gooding makes a good point,

Other things that can cause slowness:

  • 1/2 duplex connections
  • Not disabling LRO on VMWare
  • Circuit oversubscription

Have you checked any of these?

Please help us make the communities better. Rate helpful posts!

View solution in original post

Highlighted

Thanks all for your responses.

I think LRO is the cause of my issue.  It is enabled right now.  Since my VM environment is mixed with Windows, i can't disable it right now.  I am planning on making changes next weekend.  I will post my results once i make the change.

Thanks all for your help.

Highlighted

Not even a somewhat helpful point there guy?

Highlighted

My VM guy talked to VMware support and they recommended not to disable LRO.
We are running bunch of Windows servers with this VM server. Do you know if disabling LRO feature is going affect the Windows server performance?
Thanks

Sent from Cisco Technical Support iPhone App

Highlighted

Vatter,

Cisco recommends you disable it for all UC on VMWare installations.  See: http://docwiki.cisco.com/wiki/Disable_LRO

If your UC deployment has abysmal TCP performance that is why.  As a general rule I disable LRO on ESXi 4.1 and 5.0 to prevent this from occuring.

Also remember to upgrade your VMWare tools on your UC servers after installing them into VMWare.

- Steven

Please help us make the communities better. Rate helpful posts!
Highlighted

Are you running any windows servers with this VM?

Sent from Cisco Technical Support iPhone App

Highlighted

Vatter,

Yes I have run MS Windows 2003/2008 servers with LRO disabled.  Namely the AQM, WFM, and CUEAC/CUBAC servers.  There have been no issues I can recall when using these products.

If you are running non-UC Windows servers with Cisco UC on the same ESXi instance be sure to check the Cisco docwiki site to ensure the applications are fully supported in that mode.  Also remember that over-subscribing CPU cores, RAM, and disk I/O will likely cause issues. TAC will very likely recommend you stop the practice before offering detailed support. 

This page is a good place to start:

http://docwiki.cisco.com/wiki/Before_You_Buy_or_Deploy_-_Considerations_for_Design_and_Procurement

-Steven

Please help us make the communities better. Rate helpful posts!
Highlighted

Thanks for the information.. We are going to try to disable the feature this weekend.


Sent from Cisco Technical Support iPhone App

Highlighted

We disabled LRO and I even rebuilt the server but same problem. Anybody have any other ideas?

Sent from Cisco Technical Support iPhone App