AnsweredAssumed Answered

Creating a large image server cache on linux

Question asked by kent327 on Dec 5, 2014
Latest reply on Dec 5, 2014 by kent327

I'm trying to accomplish the title of this discussion. I've run into so many problems I'm not sure what to ask.

 

General info: Caching about 400 GB of MrSids in a Mosaic Dataset (ArcGIS Server estimates approximately 900 GB of cache)

 

1. We have a cluster of 4 linux servers creating tiles (Administered via ArcCatalog on Windows). During creation temp files are being written to each server. One server always takes about 3X more temp files than the other 3. During one attempt one server filled up from the temp files (the cache is being written to a shared drive) and the service crashed. Upon cleaning out the temp files and restarting the service the cache status page reports that tile generation is in progress, but the percent completion never progresses (also no job shows up in the jobs status section of the status page).

 

2. We tried doing chunks of levels, letting them finish, then deleting the temp files and starting a new chunk of levels. This seemed to be working but I tried to stop caching at some point during level 18 and we fell into the same loop where the cache status page reports that tile generation is in progress, but the percent completion never progresses.

 

3. I can't export the cache we have already generated. The export tool won't accept any of the linux machine file paths as it says cache path is not accessible. I'm just using the paths we are already writing to though.

 

The only thing I have found via searches that has worked to get out of the tile generation with no progress loop is delete the service, delete the cache, create a new service with new name, and start over. This isn't the best option when the process is going to take 10+ days.

 

Any suggestions would be greatly appreciated.

 

Thanks.

Outcomes