Changeset - 12246b166181
[Not reviewed]
default
0 2 0
Nathan Brink (binki) - 15 years ago 2010-08-08 00:19:54
ohnobinki@ohnopublishing.net
Provide a basic model for job distribution.
2 files changed with 12 insertions and 1 deletions:
0 comments (0 inline, 0 general)
doc/architecture.txt
Show inline comments
 
@@ -20,24 +20,29 @@ Concepts:
 
      distren://<servername>/job/<jobid>
 
  - job packaging: Currently, the most creative way of dealing with jobs I have at the moment is
 
      storing all of the data necessary to render the job in a tarball. This tarball can be
 
      treated as a directory of a normal filesystem. A job's directory would contain at least one
 
      file called ``distrenjob.xml'' which provides information necessary to render the job. This
 
      will be an XML file rather than a rigidly defined binary format because XML supports
 
      arbitrary data storage out of the box. This should allow different rendering backends to
 
      store the extra information that is specific to that backend.
 
  - post-render rebundling: Again, to make things simpler, the server where a job is rendered
 
      shall be responsible for collecting the individual completed frames and collecting them
 
      into a tarball. This tarball may be called for using a client. This tarball is a bundle
 
      of completed frames and will exclude the seed tarball.
 
  - job announcement: When a job is created, the server it was submitted to shall make a
 
      global announcement that the job is available. All servers should record this announcement
 
      All servers will maintain a list of globally available jobs. When a server becomes available
 
      to render a frame, it will scan the lists of announcements it has received and then ask
 
      the originating server for a frame.
 

	
 
- frame: Represents a distinct, and hopefully small, unit of work that a renderer backend can perform.
 
  - frame identification: Like a job numeral is useless without a qualifying servername, a 
 
      frame's identification number would be useless without an accompanying job identifier. The
 
      numeric value of a frame identification value must be unique to that job. There are no
 
      restrictions of ordering of frame numbers except that they are not to be negative. There
 
      need be no sequencing of frame numbers either. Thus, a frame identification URL would look
 
      like distren://<servername>/job/<jobid>/frame/<frameid>
 
  - size: A frame hopefully represents a smaller unit of work in terms of the rendering
 
      back-end's capabilities. For example, POV-Ray's CLI can initiate and carry out the rendering
 
      of an entire animation. But distren would hopefully be able to provide a clean method for
 
      rendering each individual frame and bringing the resulting set of files back to the user
doc/distrenjob.xml.example
Show inline comments
 
@@ -35,25 +35,31 @@
 
	/> tag is a better example, as povray's .ini file will specify
 
	the information necessary for a sequence of frames to be
 
	rendered and we need to actually intercept/override that
 
	information.
 
    -->
 
    <resolution
 
       width="800"
 
       height="600"
 
       />
 

	
 
    <!--
 
	Currently, for simplicity, there will be a <frame /> tag for
 
	every frame to be rendered.
 
	every frame to be rendered. However, only the server from
 
	which the job originates will see every such frame.
 
	
 
	When a server requests a frame, it shall receive a full
 
	distrenjob.xml that contains only one frame. It will thus have
 
	enough information to render that one frame and return its
 
	results to the originating server.
 
    -->
 
    <frames>
 
      <frame name="distren://ohnopub.net/job/36/frame/23" />
 

	
 
      <!--
 
	  As described in architecture.txt, we allow frame
 
	  dependencies.
 
      -->
 
      <frame name="distren://ohnopub.net/job/36/frame/22">
 
	<dep name="distren://ohnopub.net/job/36/frame/23" />
 
	<dep name="distren://ohnopub.net/job/36/frame/24" />
 
      </frame>
0 comments (0 inline, 0 general)