Skip to content

Latest commit

 

History

History
57 lines (48 loc) · 2.97 KB

README.md

File metadata and controls

57 lines (48 loc) · 2.97 KB
Raw

FileFlowFaster

It moves files, and it does it well.

Features:

  • WebUI for monitoring transfers(cancelation,reordering etc)
  • Transcoding using ffmpeg for video files
  • sender receiver transcode functionality that allows for transferring video as a compressed format and storing it at either end in a more editable format eg: IMX30( 30mb/s intra only sd footage) -> Streamed as h264 -> IMX30 |Sender | data streamed via tcp | recieved and re-encoded
  • sending files via ftp
  • limiting concurrent transfers in a kind of heirachy
  eg:       total(6)
           /       \
   outbound(4)     inbound(4)
      /            /        \
   NAS(3)    Backups(1)    VidServer(4)

Each group's number represents the max number of concurrent transfers of that group. In this example: - The total can never exceed 6. Maybe you just don't want to have to keep an eye on more transfers than this - outbound and inbound are sending and receiving over the internet. Say you have 100^/100v internet. So you decide you can do 4 concurrently each way - NAS has a lot of smaller files you are sending out. You want a few concurrent streams to minimise speed loss from switching the many files - Backups Automatically copy some kind of backup to a local storage device. Its one huge file, so there is no reason to run more than one transfer at once. - VidServer you want to copy video from a very high quality remote storage to you but in a lower resolution. A single encode doesn't max out your cpu, so you want to be able to run a few at once

  • Maybe more stuff at some point.

Features that do not exist:

  • Good documentation. It's a big complex machine, many parts still change.

How it all works

The basic gist is as follows:

--Do some clientmanager connection and initialisation stuff --

Main loop:

The project that primarily deals with it is specified in() eg: (project)

  1. (Watcher) Files get found by the watcher
  2. (TransferClient) Each file gets given to the scheduler where a job is made and added to the jobdb
  3. (JobDB) The jobdb responds to various events and shuffles jobs about. Once that job is ready to run (The file has stopped growing, and it has all its needed tokens (Each job needs a token from every group it is a part of. Each group starts with a number of tokens equal to the max jobs it allows. Once a job takes a token it is unavailable till it is returned when the job is complete.)) (The events are:
    1. New job
    2. Job finishes (deleted, cancelled success, failure, whatever)
    3. Reordering) The jobDB does nothing except when one of those events occurs. IT has no internal loop or anything like that.
  4. (JoBD) The Job Is run, which starts the mover code. (Mover) This may mean starting ffmpeg, tcp,ftp or any other process that is part of the jobs movement
  5. (TransferClient) job finishes cancels etc
  6. (JobDB) tokens are released