Right now the data that needs to be stored for multipart uploads are stored in various maps in memory. If the S3 API is restarted, that status of the uploads is lost. Consider reorganizing this to use sqlite for persistent state.
One caveat to this is that we currently store connection/stream objects in this memory which can't directly be stored in sqlite.
Right now the data that needs to be stored for multipart uploads are stored in various maps in memory. If the S3 API is restarted, that status of the uploads is lost. Consider reorganizing this to use sqlite for persistent state.
One caveat to this is that we currently store connection/stream objects in this memory which can't directly be stored in sqlite.