Back in January our internal GIS server had a hard drive failure and after upgrading to ArcGIS Server 10.5 we ran into a host of issues. More About that here. After upgrading SQL Server to 2014 we could not register the SQL Server data connection to the ArcGIS Server Data Store. The bottom line is that we were unable to publish from it and kept getting error 001487 “packaging succeeded but publishing failed”
Issues we ran into:
In addition to not being able to register or publish from the SQL serer instance we also could not republish from our file geodatabases which were in the data store without changing the name of the MXD. We installed the new publishing patch nothing happened, We messed with config files, still did not work… It was a real mess and I was at my wits end. Then I got another ESRI tech and he suggested installing ArcGIS Server on another machine.
I was like where am I going to get such a machine it’s not like I can just spin up another instance… we’re local government… I was seeing red at this point but he was patient and I remembered that I had a little mini server on my desktop that I could use aka my laptop.
So I logged in to our customer service portal grabbed out fail-over license and installed ArcGIS Serer 10.5 on my laptop. It took less than 20 minutes to set the whole thing up and I was able to register the SQL Server connections to the data store successfully. We popped on to the server and copied over the connection files from the laptop to the server and was able to get them registered to the ArcGIS Server data-store on the Server (woo hoo!) I was finally able to publish the data from the SQL Server instance and was happy as pie.
The other issue is that we had quite a few services running with their minimum instances set to 1 and above and with the number of services we have published (> 100) on the server they were eating up resources while trying to publish so we dropped quite a few of them to 0 and is noticing a big difference in the amount of time it takes to publish and the issues with publishing from the file geodatabases seems to have been abated. Now we’re looking at tweaking the pooling on the map services and trimming the number of services we have published.