Metadata migration from Windows to Linux

[Abhijit]

Hello Everyone,

We are trying to migrate our Metadata from Windows to linux/unix. We have SAS9.3 on our site. I took a Metadata backup from the Windows environment, brought over the backup files to Unix environment & did the restore. The restore was successful but it is opening in "read-only" mode? Cant do any updated.

Is there any additional configuration that we need to perform like updating "Server Name"/ "Credentials", etc which will allow the Metadata to open in normal mode?

Thanks

[PaulHomes]

For the underlying reason I would look in the SAS metadata server log for error and warning messages.

Having said that however, you will run into all sorts of problem trying to use a simple metadata backup to migrate content from one environment to another with different operating system types. All your directory paths will need updating, as will host names and possibly port numbers, any stored credentials, and user id formats to name a few.

I would strongly recommend you use a supported method such as promotion (install-new-and-import using exported .SPK metadata packages) or migration using the SAS Migration Utility (SMU) to create a package from the old environment and use it to install a new environment. In your case, since your are changing operating systems, you would need to look at the promotion technique.

For more information have a read of the SAS® 9.4 Intelligence Platform :: SAS(R) 9.4 Intelligence Platform: Migration Guide (or older version as appropriate depending on which version you are migrating to).

If you're not 100% confident, I'd also suggest getting the help of SAS Professional Services, or a local SAS partner in your area, as they have a wealth of experience in this area.

[ShelleySessoms]

I have moved this to the Admin and Deployment Community, in hopes of generating additional responses for you.

[Abhijit]

I have used the below command to create a migrationpackage

smu93_32 -properties "C:\smu93\smu.properties" -replace

I got the below files & I brought that over to Unix environment:

-rw-rw-r--   1 u35162707 dpitsdev     366 Mar 26 07:40 manifest.properties

drwxrwsr-x   2 u35162707 dpitsdev       3 Mar 26 07:40 migration-schemas

drwxrwsr-x 194 u35162707 dpitsdev     202 Mar 26 07:40 SASMDS93.dwp-asd.gov.uk

Does anyone know how I can deploy this migrated package on the Unix box?

[PaulHomes]

As I mentioned, "In your case, since your are changing operating systems, you would need to look at the promotion technique.". If you read through the migration guide it compares the migration and promotion approaches. Migration uses SMU, whereas promotion uses the approach of installing new and then promoting using .SPK packages. SMU, among other things, requires the same operating system family. In your case you are moving from Windows to Linux, so you will need to use the promotion approach. If you read through the migration guide (for the SAS version you are moving to) it provides a lot more information and guidance.

[DavidStern]

Hi Abhijit,

I'm amazed that you were able to restore a metadata server backup taken on a Windows SAS Deployment on one machine, on a Linux SAS deployment on another machine. I would not expect that to work.

Paul is right. It's not possible to use a migration package created by the SAS Migration Utility on one operating system (Windows, in your case) with the SAS Deployment Wizard on an operating system in a different operating system family (Linux, in your case).

• If you're planning to deploy SAS 9.3, see the bottom of this page: SAS(R) 9.3 Intelligence Platform: Migration Guide, beginning "the same operating system family".
• If you're planning to deploy SAS 9.4, see the bottom of this page: SAS(R) 9.4 Intelligence Platform: Migration Guide, beginning "the same operating system family".

The easiest option available to you if you wish to upgrade your application content from a Windows SAS deployment to a Linux SAS deployment will be the promotion method Paul mentioned, moving your content in SAS Package (.spk) files. If you haven't already deployed it, start by deploying SAS on your Linux environment using the SAS Deployment Wizard, but without ticking the 'Perform a migration' checkbox to provide it with a migration package (because you don't have a migration package created on another Linux or Unix environment). Follow all the steps to get it deployed, configured, validated and working properly. (SAS Professional Services, or one of our partners can help you with this if you need it).

1. First perform an audit of the content you have in your existing SAS deployment - identify all the libraries, tables, jobs, users, groups, transforms, stored processes, cubes etc that you want to move to your Linux SAS deployment. Also, audit any customizations made to your SAS configuration files on Windows, and decide whether you need to make similar customisations on your Linux deployment.
2. Apply the customisations you want to reproduce from your Windows Environment to your Linux environment, adjusting as necessary for the different operating system.
3. Then using SAS Management Console (or SAS Data Integration Studio if you have it) in your source Windows environment, right-click the root SAS Folders folder and use "Export SAS Package files" to export a package containing your metadata (sometimes including some closely-related content from the filesystem, WebDAV or table store, depending on kinds of metadata what you're exporting). Always look through the log file created by the Export SAS Package wizard, and pay special attention to any warnings or errors.
4. Then move those .spk files to your Linux environment.
5. Then in SAS Management Console, right-click the root SAS Folders folder and use "Import SAS Package" to import the content. The Import SAS Package wizard will prompt you to provide new values for associations that the metadata objects you are importing have with other metadata objects already in your new deployment, or to provide the new (Linux-style) file and folder paths for libraries, external files and so on. Always look through the log file created by the Import SAS Package wizard, and pay special attention to any warnings or errors.
6. Audit the content you've now moved to your Linux deployment. Ensure it is all present, that it looks correct to a quick inspection, and that it works as intended (with no more errors than it has in your Windows environment).

When you're converting from Windows to Linux, remember that any hostnames, folder or file paths and other platform-specific references embedded in your code may need to be updated.

And as Paul already said, there's no substitute for reading all the documentation before you start - it is packed with useful information which can help you avoid spending time on attempting to do something which won't work. Best of luck and do let us know how you get on!