The issue with the redwood upload is that the connection between the client and the server times out at the last step when calling the final upload call. This results in the client retrying the finishing call, which causes the redwood server to initiate the deletion of the key parts. Since they are no longer there, this results in a 404 error from AWS. Since deletion of these part keys takes more time the larger the file, there is a point in which 60 seconds (the default timeout for the redwood server) is not enough to wait for all the part keys to be deleted. A workaround for now is to just increase the timeout in the config file for the server. It should be noted that if there is a front facing server like NGINX or Apache, the gateway timeout should also be increased. Otherwise the Redwood server will not timeout but the gateway will, resulting in the same error.
It would be good to investigate if the http connection client has a setting for keeping alive the connection (e.g. sending a heartbeat, etc).
The issue with the redwood upload is that the connection between the client and the server times out at the last step when calling the final upload call. This results in the client retrying the finishing call, which causes the redwood server to initiate the deletion of the key parts. Since they are no longer there, this results in a 404 error from AWS. Since deletion of these part keys takes more time the larger the file, there is a point in which 60 seconds (the default timeout for the redwood server) is not enough to wait for all the part keys to be deleted. A workaround for now is to just increase the timeout in the config file for the server. It should be noted that if there is a front facing server like NGINX or Apache, the gateway timeout should also be increased. Otherwise the Redwood server will not timeout but the gateway will, resulting in the same error.
It would be good to investigate if the http connection client has a setting for keeping alive the connection (e.g. sending a heartbeat, etc).