CUCM 6.1.3 Cluster NodeID disordered, DMA ?

[Nguyen Quang Son]

Hi everybody,

I'm taking over a CUCM 6.1.3 cluster and when I did a "run sql select ccm, nodeid from processnode", I got this

name     nodeid

pub            2

sub4          3

sub2          4

sub3          5

sub1          6

Informed that the name convention followed the installation order, I expected to get

name     nodeid

pub            2

sub1          3

sub2          4

sub3          5

sub4          6

The question is: what could disorder my servers' nodeid ? Do changing subscriber's IP address have anything to do with that?

A friend told me this could be the result of a DMA migration. Can someone please confirm this ? A Cisco link is really welcome.

Thank you in advance, pros,

Best regards,

[Cedric Van Labeke]

It can really be the order in which the servers got installed. Especially because there are no "jumps" where some node is is missing.

Basically, anytime you add a server on ccmadmin pages System > Server, a new node will be written into process node table.

Please see below example from my lab:

admin:run sql select * from processnode
name                              nodeid

==================  ======
10.48.79.35                     2             (pub)
10.48.79.36                     4             (sub)

Notice there's not even node 3 here :-)

Now deleted the sub from System > Server

admin:run sql select * from processnode
  name                              nodeid

==================  ======

10.48.79.35                      2          (pub)

admin:

Now added the sub again on System > Server

admin:run sql select * from processnode
  name                              nodeid

==================  ======

10.48.79.35                     2           (pub)
10.48.79.36                     5           (sub)
admin:

Did they add/remove servers throughout the lifecycle of the cluster?

[Nguyen Quang Son]

Hi Cedric,

Thank you very much for your test. Again the NodeID distribution is quite random.

- Why there isn't a nodeID 3 in your Lab, I suppose you did a add/remove before?

- Why 10.48.79.36 does not take back it nodeID? I suppose it's the normal behavior when you remove then add back a server. But why it didn't take the 3?

There was no add/remove server on the cluster.

Have you ever experience a DMA migration. I suspect that would be the root cause.

Regards,

[Cedric Van Labeke]

Hi Nguyen,

I've done many DMA installs and troubleshooted even more of them I think :-) No, I haven't seen this being caused by DMA yet.

If you search CCO / or Google for the Database Dictionary of CallManager, and you look at the description of that table and its field, you will notice that the node-id field is marked as identity (Unique, Indexed, Not Modifiable) similar to the pkid (primary key) whereas another value such as name, is not. This is most likely why it is not taking it back / reusing it.

My bet is that the servers got installed in the order of the node-id you see below.

May I ask why the node-id not being what you expect it to be is a problem?

How is it impacting you or your customer?

Thanks,

Cedric

[Nguyen Quang Son]

Hi Cedric,

My bet is that the servers got installed in the order of the node-id you see below.

--> I got the confirmation from the tech guy that installed the system. This is not the same order as the node-id. In fact. He started with Sub1 and TFTP1.

May I ask why the node-id not being what you expect it to be is a problem?

How is it impacting you or your customer?

--> It 's not really a problem to us and it does not impact our customer (so far). Just want to stay clear . Moreover, it may cause confuse for trace reading.

Thanks.