Hi @j2,
I would set a limit to the number of loops (just for testing purposes). This should allow you to verify the same behavior on Desktop and Cloud. Keep in mind too that FME Cloud is on Linux and playing with file paths can be a real pain when testing between windows and Linux because of the difference in paths.
Hi @j2,
I would set a limit to the number of loops (just for testing purposes). This should allow you to verify the same behavior on Desktop and Cloud. Keep in mind too that FME Cloud is on Linux and playing with file paths can be a real pain when testing between windows and Linux because of the difference in paths.
Hi virtualcitymatt,
Thanks for the reply.
At first I thought it was a problem with the looping transformer but it seems it might be more related to what you're saying.
When running on Desktop I use the File/Directory reader and choose a path like
C:\\Projects\\testproject\\subfolders where "testproject" becomes level 0 or the root of the folder.
On Cloud it starts an infinite loop if I select a path like:
$(FME_SHAREDRESOURCE_DATA)/testproject/subfolders
However it does make the first loop if I change it to:
$(FME_SHAREDRESOURCE_DATA)/testproject/testproject/subfolders
Then it goes into the infinite loop again.
Edit:
I now tested the first version of the workflow without the loop on the 18528 version of Cloud and it doesn't work either....Seems like File/Directory reader doesn't really work on Server/Cloud?
Hi @j2,
can you share a simple workspace (without looping) and complete log files that demonstrate & reproduce the problem on FME Server 18528?
There is no indication that the Directory and File Pathnames is not working on FME Server or FME Cloud right now, but I am happy to test your scenario if you can share a simple workflow.
Hi @j2,
can you share a simple workspace (without looping) and complete log files that demonstrate & reproduce the problem on FME Server 18528?
There is no indication that the Directory and File Pathnames is not working on FME Server or FME Cloud right now, but I am happy to test your scenario if you can share a simple workflow.
Hi GerhardAtSafe,
Thank you for your reply.
I just tried both workspaces on FME Server 18528 on a Windows machine and both worked perfectly. So it seems the problem is related to what virtualcitymatt is talking about regarding the file paths? Are there any general guidelines for using Directory and File Pathnames when it's running on a Linux machine?
I will make a simpler version of the workflow to share.
Hi virtualcitymatt,
Thanks for the reply.
At first I thought it was a problem with the looping transformer but it seems it might be more related to what you're saying.
When running on Desktop I use the File/Directory reader and choose a path like
C:\\Projects\\testproject\\subfolders where "testproject" becomes level 0 or the root of the folder.
On Cloud it starts an infinite loop if I select a path like:
$(FME_SHAREDRESOURCE_DATA)/testproject/subfolders
However it does make the first loop if I change it to:
$(FME_SHAREDRESOURCE_DATA)/testproject/testproject/subfolders
Then it goes into the infinite loop again.
Edit:
I now tested the first version of the workflow without the loop on the 18528 version of Cloud and it doesn't work either....Seems like File/Directory reader doesn't really work on Server/Cloud?
Hmmm, I use that on FME cloud often without issues (except me), I suppose there could be an issue with it on the version of FME on Linux but unlikely. Check with Loggers in your workflow to see what the path reader is actually reading. Also put loggers before the reader (if it's a feature reader). Loggers are your friend here when working on FME Server. Also you need to make sure that if you're using the pathname which is read that you use the _path_unix rather than the _path_windows attributes.
Ultimately it had nothing to do with the transformer but it seems paths are interpreted marginally different on Cloud compared to Windows Desktop/Server?
I ended up using a generic reader which reads zip files instead which uploads and extracts the files to Cloud/Server and then run the HTTP looper.
Thanks for the hints and replies!
Ultimately it had nothing to do with the transformer but it seems paths are interpreted marginally different on Cloud compared to Windows Desktop/Server?
I ended up using a generic reader which reads zip files instead which uploads and extracts the files to Cloud/Server and then run the HTTP looper.
Thanks for the hints and replies!
Hi @j2,
glad to hear you found a solution!
The main difference on FME Cloud is that the underlying OS is Linux and therefore unix paths are used instead of windows paths. Unix paths use / as separator and windows paths use \\. The Directory and File Pathnames reader returns both options and therefore you would have to switch to the path_unix attribute before you publish your workspace to FME Cloud. Another common issue when dealing with pathnames is spaces and special characters. Using quotes around pathnames can help with this.
