Skip to main content
  • EN

Hello everyone,

For one year, I’ve been running different workbenches daily (Windows Scheduler), reading from MSSQL tables to write into various hosted feature layers, used in various dashboards. I noticed an unusual consumption of the storage on our GIS server, coming from the “pgdata” folder of Arcgis Entreprise, so the underlying postgresql database (up to 20go additional Gb per month, while the size of all my hosted features stay constant at around 1gb only in total). After several tehnical sessions with ESRI, there is no doubts that this is caused by my FME workflows. 

Those workbenches are far to be optimised (as I was only starting to use FME) but they are working fine. It was easy to truncate all features to insert new features, sometimes up to 150 000 features daily, for one layer (see below, simplified version). The inserted features are polygons, from the same source hosted feature layer, daily joined with most recent MSSQL data. NB: Objectid is removed with “attribute manager”. Beginning of 2024, I only inserted data for 2024 and not for previous years, so I expected my pgdata folder to decrease in size, which was not the case. Moreover, the performances of the dashboards decrease a bit over time. I suspect that the truncating is not working as intended and that features accumulate over time (ghost indices ?). Have you already encoutered that issue ?

I understand that using “change detector” to assign fme_operation is the correct practise and should prevent future problems, but I’m not sure if this will fix the the past accumulation issues. Is there a way to really overwrite an hosted feature layer with FME ? Am I missing something ?

Thank you in advance,

Arnaud

 

 

 

 

I imagine the size is due to the DB not clearing out old indices (as you alluded to), that is an Esri thing to worry about as Esri (Enterprise) manages that DB. It could just be a case of not hitting the default cleanup limits and not triggering internal process to clean up old things.

In regards to change detector, yes, that is a good approach. Essentially you read in the ‘new’ data and the ‘old’ data. Setup the change detector to compare based on various conditions and it will output features that are new, changed, deleted and unchanged. You can then action each of those group appropriately.

The help doc is pretty good - https://docs.safe.com/fme/html/FME-Form-Documentation/FME-Transformers/Transformers/changedetector.htm

And i found this webinar that might help as well - https://engage.safe.com/webinars/managing-changing-data-with-fme/

Reply


Community Stats

29,744
Posts
113,078
Replies
37,403
Members
Terms & ConditionsCookie settings