We run the Manage tile cache tool often to create caches.
The data source is image service and the target cache can be 1-5Gb in size. The process can take up to 2-4 hours.
The process ends successfully.
When we check the cache we found that is some levels in some area we see white screen.
Checking the disk sizes some levels sizes are smaller than expected.
Checking the directories on-disk I found that for the levels with problems the XX.done file does not exists for each XX.bandle file.
Yet the tool itself and the server of the source image server does not show any errors.
Trying to use the manage tile cache again with the "recreate empty tiles" help some time but not always.
I hope this helps.
I am using a 40 core machine and use 50% so 20 threads are open.
It is a network disk but the machine itself is virtual so I am not sure it makes different
Sometimes from 500 to 2000000, each level is twice the previous level. But sometime the 500 and 1000 is fine while the 2000 is not...
I am aware of this option but all the operation is automatically and here I have to do manual work (:
I have faced the same issue when I was generating tiles using AGOL scales and my machine was a 48 core. The tiles would generate till 15th level and then it will generate partial tiles.
I had to eventually re-configure the server since the issue was with the configuration and then we increased the Maximum number of instances per machine to 42 and it did the trick.
Maybe try to increase the the number of instances and see if it resolves the issue, else I'd recommend to log a ticket with https://community.esri.com/groups/technical-support?sr=search&searchId=fca4382e-f06e-4496-b8a8-a01a5...
We have a virtual machine with 40 cores that call an image server on another machine with 12 cores.
Did you just run the local machine with 90% and not 50% or are you talking about the image service machine?
Is it possible that the number of cores on the client should not be bigger than the number of instances for the image server machine?
I had just the server placed in the data center and no image server.
The caching happens on the server the server so I'd assume the number of cores should be more on the parent server compared to the client server.
We have the same issue.
As for now, two reasons found:
1) Mosaic dataset has overviews with a broken path to the file.
- Run Export Mosaic Dataset Path tool with Export Mode: Broken.
- Delete the overviews.
- Rerun Manage Tile Cache tool
2) Manage Tile Cache tool runs with Area of Interest
It is a known bug fixed in version 10.5.1.