New items in feed source are not detected (or imported)

[izmeez]

When a feeds importer is created or edited the Preview import function allows browsing to select a file and then user can click the "Preview import" button.

However, when the selected file is then updated and the "Preview import" button is clicked again it does not use the updated file. It is as though the file has been cached. This should not happen as a normal workflow may be to preview the import, make adjustments to the file and preview the import again to confirm it is working. A solution may be for the preview import function to reload the file before preview.

[jenlampton]

I am running into this also, but in my case it is not just the preview that is not aware the source file has changed. The actual import behavior matches the preview behavior. My manual imports and cron imports also fail to import any of the new items in the feed.

The only way I can get my importer to create new node is:

1. Go to the list of files on the system admin/content/files
2. Look for any file name that starts with FeedsHTTPFetcherResult
3. Delete all these files (they are all be marked as temporary)
4. Go to the importer page import/importer_name
5. Click Import

I have also disabled the Cache HTTP result of request under the HTTP Fetcher settings -> Advanced , tinking maybe this was what was causing this problematic cache of the source data, but even with this checkbox un-checked, the temporary files appear in the list of files, prevent new information from coming into the system, and are not deleted after the import runs.

[jenlampton]

I just imported an older database that has 4 files named FeedsHTTPFetcherResult1770341463 from over a month ago, ran the importer, and no new file was created.

I ran cron, which deleted the old temporary files as expected, and also ran the importer again as expected, and now there are 4 new files named FeedsHTTPFetcherResult1774466373.

My first question is, why are there four?
My second question is, why are these used instead of pulling a fresh copy during each run?

I found this code comment for FeedsFetcherResult->getFilePath():

File will be deleted after BACKDROP_MAXIMUM_TEMP_FILE_AGE.

This value is 6 hours. My importer is supposed to run every 3 hours.
That means that every second run it's going to be missing new items from the feed if it always uses a local file (if it exists) instead of fetching from the feed.

FeedsFetcherResult->getFilePath() doesn't seem to be called during my import. I also tried getFile() but that wasn't called either. I even tried a state_set('feeds_reschedule', TRUE); to make sure it would try to import, but now I'm not sure what is even saving these files.