So, nothing really formal, but a list of things to check as you complete your "merged server" process, which all of us dual NPM/NCM server people get to do! Some of these may be addressed in 7.3, but doubt it will hurt to check!
Pre-Upgrade:
Make a copy of the following directories:
<install area>\Orion\NCM\DeviceTypes\*.*
<install area>\Orion\NCM\Scripts\*.*
<install area>\Orion\NCM\Reports\*.*
<install area>\Orion\NCM\Jobs\*.*
Document all of your configured NCM jobs from the Win32 client - Scheduled Jobs area. This is good not only for the upgrade, but for your own backup documentation.
Upgrade
Before you start your upgrade - go to your NCM server, open Windows Task Scheduler and disable all of your NCM jobs. Do not delete them, disable them.
Ensure you run the NCM additional poller install file on each poller. (Following your main server upgrade)
Post Upgrade - Using your pollers for NCM
First - before you can use all of your pollers for NCM operations - the devices in your network must be configured for the access. Usually SSH. Assuming your pollers are set to go, and you have checked the option of "Additional Polling Engine" under Polling Configuration:
1) From your backup copy - push out to each poller the DeviceTypes, Scripts and Reports folders to the appropriate area on each poller. You may merge these folders to the existing folders created in the install
2) Folder permissions: If any of your jobs delivered output to a folder on a server - ALL POLLERS now need that access. Easiest way I found was to give the share and the folder permissions to the individual pollers (PollerMachineName$). Any directory not set as such, go modify now. You could run the Job Engines under a service account with permission to that folder, but would prefer to not poke at SolarWinds permissions. Easier / Safer to just leave the services alone, and modify your shares with the machine accounts. This will allow scheduled jobs to deliver files to remote folders.
3) Go to "Settings" - "NCM Settings" - SMTP Server. Put the appropriate SMTP server information in. This will save you time later.
4) Go to "Settings" - "NCM Settings" - Email Notification. Put your default information in here. This will save you time later.
5) Go to "Settings" - "NCM Settings" - SCP Server. If you use this functionality - enter in your correct data here. Ensure the SCP service is started on all target machines.
6) Go to the Config tab, Jobs page. This is where your previous NCM jobs will be, and you create new ones. If all you have are example jobs - then they did not migrate. You will have to use your documentation to recreate all your jobs. Didn't get everything documented? No problem.
A) Open the old NCM server Task Scheduler. Look for the disabled job that you need to recreate. Task Scheduler will give you the run times, and a Job file it runs. Format will look like "Job-12345"
B) On the Jobs folder you backed up, find that job file. Open in Word pad, and it will list out the rest of the details you need.
7) Once you have created your jobs - WAIT. While it probably doesn't take this long - go have a cup of coffee. Go to lunch / dinner. The jobs must propagate out to the pollers, so they know they have work for their nodes. Lack of patience in creating a job, leads to the job not being out on all the pollers, which leads to time frustratingly trying to figure out why some worked, and some didn't. (Fear leads to anger...anger to hate....you suffering)
8) If you keep a copy for the device Config Downloads, make sure your path, and the format is correct. You should only have to change the format, if you did so from the default setting. Check anyway!
9)Test your jobs. Chances are, if a device fails to download a config, it is one of two things: Poller does not have access permissions to that device (cannot log in) or cannot find a correct template to use (Could not find device template <install area>\NCM\DeviceType\Some-Name) If you get an error saying you cannot write to an export location - check your SHARE and FOLDER permissions.
Edit 1:
Exporting Results to a File
Until corrected in a future release - it is NOT possible to have an NCM job, which will run on two or more pollers, write to a single file on a network share. You will see your job start, and remain "running" forever, even though the job log will indicate it has completed. If you have each poller write to it's own local drive, this does not affect you.
The cause: The job will start on each poller at slightly different times. The first poller to start writing to the destination file wins. The others will not write to the file.
The work-around for this, if you require a single file on a network share:
1) Use the format of \\ServerName\ExportArea\${HostName}JobName.txt
2) Write a Windows Task Scheduler to run later on that will cat (concatenate for those non-*nix people) the file to your single destination file. Probably easiest to do it on the destination server, if possible.
Feel free to add suggestions and lessons learned. I will likely add in a few more as I think of them later.