WEBGISDR Restore successful, but no REST services and Content is blank

[MikeSchonlau]

We were able to successfully run the WEBGISDR Restore process from an ArcGIS Enterprise HA 11.2 environment to a Single Machine 11.2 environment. However, even though the WEBGIDR --import process was successful and we can validate our Data Store and Hosting Server, we not seeing any REST services and all of our content is blank in Portal. Users and Groups came over fine. Any thoughts? 

This is an AWS-hosted ArcGIS Enterprise 11.2 environment. Been following this blog as a guide along with the documentation.

[NicolasGIS]

Seems more like a Portal for ArcGIS issue to me than datastores. What is not displayed correctly is the result of the following URL: /sharing/rest/search. Did you maybe try to reindex (portaladmin/system/indexer/Reindex ?). How does your Portal for ArcGIS content folder look like ? Was it successfully restored (ie: can browse your items and find the thumbnails and so on) ?

Disclamer: just trying to help by throwing ideas, not working for ESRI !

[MikeSchonlau]

Yes - the problem definitely seems to be with Portal and the REST services. It looiks like reindexing is not an option at this version based on this article. I am now more curious about the portal content directory as the old one was hosted in AWS S3 and the new one is hosted on the ArcGIS Enterprise machine. I'll take a closer look. Thanks for the advice.

[Scott_Tansley]

Things may have changed, certainly the diagrams have disappeared, but WEBGISDR used to require the same number of machines to be available on both production and DR environments.  I know that over time it became only important that the public facing URLs was available, and that you could change the private URLs for example.  However, I note you're going from a HA to a single machine DR.  The documentation for 11.2 WebGIS DR tool clearly states that the HA configurations will be maintained:

• When you restore a highly available ArcGIS Enterprise organization, the WebGISDR tool maintains high-availability settings for all components. (source)

My 'feeling' is that what you're trying to achieve is not a supported/expected workflow.

[MikeSchonlau]

Scott

Thanks for the feedback. We were concerned about that problem, but the blog post does show that HA to Single Machine is a supported operation for the WEBGISDR field. However, our existing HA environment is actually problematic anyway because we have only have one operational server (no standby). That's a long story in itself, but it is the only way we could get it working while we try to restore this to a new environment. We actually tried the WEBGISDR process to a new HA environment also, but the restore ran into problems downloading content from S3. Anyway, we are planning to try some additional restore operations as we move closer to the "desperation zone." Restoring an AWS-hosted ArcGIS Enterprise environment has been more challenging than we hoped.

[Scott_Tansley]

The infrastructure can have a big impact on a lot of these things.  ArcGIS Enterprise on a single machine is 'relatively easy' but there are so many variations in play by the time you bring in different cloud solutions and other enterprise building blocks.  Be aware that sometimes you just have to 'call it' and start a fresh.  It's a hard decision, but when migrating if you have problems in the old, then you'll have problems in the new.

[MikeSchonlau]

Agreed. I've been doing this a long time both with ESRI and AWS and this is not the first nor the last time I have considered the "fresh start" option. Looks like I'm headed that way here. Our experiences with Cloud Builder have not been great, so we're going back to building manually or maybe using a custom CloudFormation template. Not thrilled about High Availability architecture either - lots of moving parts. It's been more like the Highly Unavailable architecture so far for us. Anyway, thanks for the advice.

[Scott_Tansley]

I hear you.  I've seen issues with HA and slight nuance differences between the builders.  I have actually built my own scripts for deployment, partly PowerShell, partly 'written'.  It has allowed me to build consistent, reliable environments for all my clients.  Having all clients use a similar approach, where some are AWS, Azure, on premise or 'other' has led to efficiencies for my small business and better support for my clients.  

Frankly, I've found well architected/deployed/configured environments can achieve higher uptimes than HA, and sometimes that needs a manual build, as you suggest.

[hlindemann]

Hi @MikeSchonlau, can you have a look under "your_install_dir"\arcgisserver\config-store\services\hosted, is the hosted services reflecting here? if not then your webgisdr might not have backed up the hosted services, you can try export site under https://dns:6443/arcgis/admin it only exports ArcGIS Server, and import it to the new site.

Hope it helps 
Henry

[MikeSchonlau]

Henry

Thanks for your feedback. You are definitely on to something here. In our existing environment, our config-store uses cloud storage (AWS DynamoDb). In our restore environment, we want to use FileSystem. I checked the config-store folder on our Restore instance and none of our services are there. That is probably the root of our problem. Thanks for suggesting that. The moral of our story is that if you deploy AGE using cloudbuilder for AWS, restoring gets more complicated.