[11:33:29] Is there any case where I can get a null CSRF token ? [11:33:39] Or the CSRF token is a empty string [11:33:55] Context: https://phabricator.wikimedia.org/T345890 [11:35:50] Sohom_Datta: i don't think it's ever fully empty. but logged-out users get a single plus sign or something similar to that [11:37:30] +\ iirc (to confirm that whatever library they’re using escapes everything correctly) [11:37:48] I also don’t think an empty string should ever happen [11:40:27] So https://gitlab.wikimedia.org/cloudvps-repos/videocuttool/VideoCutTool/-/blob/master/server/controllers/router-controller.js#L12 is the source code of what we are doing to upload videos [11:41:29] It directly takes the token and plugs it into the upload API, I don't think we are losing the token somehow ? [11:42:07] Also, I can't for the life of me reproduce this, but multiple other peeps can (We've go like 5 reports of this :() [11:43:11] I wonder if the API returns +/ and that shows up as token not present ? [11:43:19] *+\ [12:55:58] I don’t see anything obviously wrong with that code, or at least nothing that I’d expect to cause this particular error message [12:56:13] though it is suspicious that the actual upload request is missing the Content-Type header compared to the token request [12:57:49] I’d try to add some more logging and see if that clears up things [13:40:16] Yeah will do that, I'll add some logging so I can get a better understanding of what's going on