[cisco-voip] building a new cluster from DMA (getting from point A to point B)

Mon Aug 9 16:13:41 EDT 2010

Thanks Wes. 

According to the v7 SRND, it's highly recommend, not required: " Cisco highly recommends configuring the publisher to point to a Stratum-1, Stratum-2, or Stratum-3 NTP server to ensure that the cluster time is synchronized with an external time source." We currently have a stratum 4 available but might be able to push a bit for stratum 3 access. 

Regarding: ws:Yep! Just re-ip your publisher. change it under system->server and then from CLI/platform. Then reboot. After that install subs. 

I'm installing the publisher from scratch as well and importing DMA during the install. The process already asked me if I wanted to modify the IP address from 10.104.20.105 to 10.104.91.105 and it did everything for me already. 

When I go into the GUI the IP address is already updated. All I have to do is change the CallManager name from uogccm105 to iptccm105. 

I think I'm good to go moving the publisher onto the live network without any issues. I did a wireshark capture and did not see any attempts to contact the old IP addresses. 

--- 
Lelio Fulgenzi, B.A. 
Senior Analyst (CCS) * University of Guelph * Guelph, Ontario N1G 2W1 
(519) 824-4120 x56354 (519) 767-1060 FAX (JNHN) 
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 
Cooking with unix is easy. You just sed it and forget it. 
- LFJ (with apologies to Mr. Popeil) 

From: "Wes Sisk" <wsisk at cisco.com> 
To: "Lelio Fulgenzi" <lelio at uoguelph.ca> 
Cc: "voyp list" <cisco-voip at puck.nether.net> 
Sent: Monday, August 9, 2010 3:52:30 PM 
Subject: Re: [cisco-voip] building a new cluster from DMA (getting from point A to point B) 

Inline again, ws. 

On Monday, August 09, 2010 3:32:40 PM, Lelio Fulgenzi <lelio at uoguelph.ca> wrote: 

Thanks Wes. 

I installed the publisher without an NTP source (it sync'ed to the hardware clock) and then when I tried to install a subscriber it choked, said it couldn't contact the NTP server on the publisher. I tried reconfiguring NTP on the publisher to the local switch, but I don't think it liked it. ws: not likely, CM needs a stratum 3(?) or better source. See SRND. Configuring stratum on router/switch is undesirable as NTP process will monitor clock drift. 

Anyways, the subscriber installed terminated abruptly after I said retry. 

I'm not sure how to make the publisher believe the switch is a valid NTP source. Any clues? I don't have to reinstall the publisher do I? 
ws: you can definitely add one from CLI. you may be able to do it from platform web pages. Just give it a valid NTP reference. You can restart NTP to expedite the sync. Use 'utils ntp status' and make sure you're not seeing '*127.127.1.0' as this indicates you're still using the internal clock. 

Regarding, the cluster itself...everything is new hardware - pub and subs. The steps you point to refer to changing the IP address on the subscribers, but they don't exist yet. This leads me to believe that changing the IP addresses on the publisher is sufficient, then I begin re-installing the subscribers with the new IP addresses. 
ws: Yep! Just re-ip your publisher. change it under system->server and then from CLI/platform. Then reboot. After that install subs. 

Perhaps it will just be easier once the publisher is built and the IP addresses are changed, to move it back to the production network, configure a valid NTP source and start building the subscribers after that? 
ws: just need valid NTP source so clocks get in sync before database tries to replicate. 

--- 
Lelio Fulgenzi, B.A. 
Senior Analyst (CCS) * University of Guelph * Guelph, Ontario N1G 2W1 
(519) 824-4120 x56354 (519) 767-1060 FAX (JNHN) 
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 
Cooking with unix is easy. You just sed it and forget it. 
- LFJ (with apologies to Mr. Popeil) 

From: "Wes Sisk" <wsisk at cisco.com> 
To: "Lelio Fulgenzi" <lelio at uoguelph.ca> 
Cc: "voyp list" <cisco-voip at puck.nether.net> 
Sent: Monday, August 9, 2010 2:48:29 PM 
Subject: Re: [cisco-voip] building a new cluster from DMA (getting from point A to point B) 

Inline, ws. 

On Monday, August 09, 2010 9:53:54 AM, Lelio Fulgenzi <lelio at uoguelph.ca> wrote: 

I'm going through building a new cluster that I'd like to run concurrently on my network with new IP addresses and CallManager names. I have always used IP addresses and would like to continue using IP addresses for now. I've installed 7.1(3a) along with a DMA import. Then I applied 7.1(5) upgrade. No problems. I did this all in an isolated switch with only the publisher and FTP server. 
ws: make sure you provide it a valid NTP reference, espcially if this standalone will become a cluster. 

Can I just go into the servers, change the IP address, then go into the CallManagers and change the names? Is it that simple? That way I wouldn't have to bother messing around with CallPark numbers and CallManager groups. If simply changing them is the answer should I change one first? 

Once I confirm there is no communications happening from the new publisher to the old cluster (via wireshark) I can then plug in the publisher into the network and start installing new subscribers. 

This is what I have now: 

publisher hostname: iptccm105 
publisher ip addr: 10.104.91.105 

(during the install, the process asked me about changing the ip address, but i can't recall if it asked me about changing the name, but i think it did) 

The old/new information is below. 

Servers: 

10.104.20.101 (OLD) -> 10.104.91.104 (NEW) 
10.104.20.103 (OLD) -> 10.104.91.103 (NEW) 
10.104.91.105 (no change) 
10.104.96.104 (OLD) -> 10.104.91.204 (NEW) 
10.104.96.106 (OLD) -> 10.104.91.205 (NEW) 
10.104.96.108 (OLD) -> 10.104.91.203 (NEW) 

CallManagers: 
uogccm101 (pointing to 10.104.20.101) -> iptccm104 (pointing to 10.104.91.104) 
uogccm103 (pointing to 10.104.20.103) -> iptccm103 (pointing to 10.104.91.103) 
uogccm104 (pointing to 10.104.96.104) -> iptccm204 (pointing to 10.104.91.204) 
uogccm105 (pointing to 10.104.91.105) -> iptccm105 (pointing to 10.104.91.105) 
uogccm106 (pointing to 10.104.96.106) -> iptccm205 (pointing to 10.104.91.205) 
uogccm108 (pointing to 10.104.96.108) -> iptccm203 (pointing to 10.104.91.203) 

ws: Are you replacing the whole cluster or trying to replace just individual nodes? If you are replacing the whole cluster then you just need follow the proper process to change server hostname and ip: 
http://www.cisco.com/en/US/docs/voice_ip_comm/cucm/install/7_1_2/ipchange/ipchg712.html 

If you are looking to replace certain nodes within the cluster then you will need to reinstall and restore those nodes from backup. This is especially true if you are attempting to replace the publisher node in an existing cluster. 

/Wes 

--- 
Lelio Fulgenzi, B.A. 
Senior Analyst (CCS) * University of Guelph * Guelph, Ontario N1G 2W1 
(519) 824-4120 x56354 (519) 767-1060 FAX (JNHN) 
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 
Cooking with unix is easy. You just sed it and forget it. 
- LFJ (with apologies to Mr. Popeil) 

_______________________________________________
cisco-voip mailing list cisco-voip at puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-voip 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/cisco-voip/attachments/20100809/ba98c73b/attachment.html>