Not really a solution, but an idea to test:

I feel like I have had this happen before, though it was awhile ago. One thing you could try, since you said hardcoding the path worked correctly, would be to try hardcoding the start of the path instead of the entire path coming from one attribute value. Something like D:\Folder\@Value(path_windows)

Thanks for the suggestion ​@liamfez - just updated the workflow, however the same error is still cropping up. Initially tried it when cached up to the Aggregator and it worked, then tried rerunning the full flow and it failed.

In addition, the reader_error attribute created contained the following message (this was the same for full path from attribute and part path from attribute).

“Failed to obtain any schemas from reader 'GEODATABASE_FILE' from 1 datasets. This may be due to invalid datasets or format accessibility issues due to licensing, dependencies, or module loading. See logfile for more information”

I’m running through some further testing… changing settings etc to see if there is anything that can force consistency in the reads.

So far, the most promising option is to Enable Cache in the FeatureReader with a timeout of 1hr. Currently i’ve run through 2 full runs successfully, but will have to see if the error crops up again and keep this chat updated.

Really interested to get thoughts on if this would likely be a long term solution or if there may be a better way?

Looking at your log again it mentions licensing; perhaps try changing the reader format to FILEGDB (open API) instead of the licensed GEODATABASE_FILE format you are currently using.

Thanks ​@liamfez. I was still getting rejections when running with the cache so this new suggestion is appreciated.

I’ve updated the readers to the FILEGDB (open API) one and so far seems to be working well. Only difference is that I need to exclude any non geometry items, which you can do in the licensed GEODATABASE_FILE format, but not the FILEGDB (open API). I got around this by exposing the fme_geometry{0} attribute and removing any “fme_no_geom” datasets. 

I’m feeling more confident with this, but will keep this thread updated in case that changes. Thanks for your help with this!

Hi All, just updating this with some further findings. I’m still getting some inconsistencies reading. I’ve added both readers in (Open API & Licensed GDB) so if it fails on 1 it will try the other. This seems to be working better, but still getting some issues. Doing a little digging in the logs and there is mention that

“FileGDB Reader: Feature class 'Chalk_Rivers_England' has shape type 'esriGeometryLine' which is unknown to FME”

Any ideas on reasons for this or ways to change this shape type for the Feature Class before reading into FME?