hepdata icon indicating copy to clipboard operation
hepdata copied to clipboard

submission: check behaviour of perform_upload_action

Open GraemeWatt opened this issue 3 years ago • 1 comments

Today there were some problems with the web pods of the Kubernetes deployment. Uploading a small test file (few KB) to the Sandbox gave a timeout error after five minutes with this message displayed: https://github.com/HEPData/hepdata/blob/fcdc4d5635b9f85754b0fcea57428202b3211b4f/hepdata/modules/records/static/js/hepdata_record_js.js#L87-L89 However, regardless of the error, it seems the code proceeded to the /consume endpoint, resulting in a record getting stuck in a "processing" state. I don't see how this could arise, since the line: https://github.com/HEPData/hepdata/blob/fcdc4d5635b9f85754b0fcea57428202b3211b4f/hepdata/modules/records/static/js/hepdata_record_js.js#L81 which executes the /consume endpoint should only be run for success but not in the case of an error. This might be some weird problem specific to the production instance, but it could be worth checking the logic of perform_upload_action in the case of a timeout error.

GraemeWatt avatar Feb 05 '21 19:02 GraemeWatt