Frame Checking

Directory browse mode

You can enable a fast mode for the function of the rrServer that searches for the output frames.

It has a large effect on a if your rrServer is running on a Windows machine, rrServer on Linux might become a bit faster.

Windows only: This options does not work for all fileservers.
Therefore you have to test it:
If you save your settings in rrConfig, then this setting has an immediate effect. 
Let the server check a finished job. If the rrServer does not find the job frames any more (Frames Done:0), revert this setting.

Hint: No matter which option you choose, it is always bad to have thousands of files in one folder. 
Use subfolders for layers and versions.

Definitions / Information about synced fileservers

There are multiple ways to synchronize data on multple sites/fileservers.

They are sometimes called caching fileserver, synced fileserver, distributed storage or replicated storage.

There are multiple different types of solutions.


A) Solution without metadata pre-sync
If a fileserver detects that new data is available, it starts to copy the data.
If you have many or large files, then it takes some time until you see all new files on your other fileserver.

B) Solution with metadata pre-sync

If a fileserver detects that new data is available, it sends the metadata (file name, size and date) of all files to the other fileserver.
Which means you can see the files within ~30 seconds (or more time depending on the solution)
Once that is done, there are again 2 different solutions:
B1) Direct Copy
    The file data is copied.
B2) On Demand
    The file data is not copied. 
    It is copied "on demand". 
    Which means once you try to open the file, the actual data is copied from the other fileserver and saved in this fileserver as well.


Note that B1 and B2 may be combined.
Example:
You specify "B1 direct copy" for some important files. 
And "B2 On Demand" for everything else.

Another Example:
Your main fileserver in your office, the second in the cloud which you use for Maya rendering.
The cloud fileserver works "B2 On Demand". 
It copies your Maya scene and texture only if you start a Maya render.
This way you save traffic and space in the cloud. 
You do not need a complete copy of all files.
But the main fileserver "B1 direct copy".
Once a cloud machine has rendered a frame, it is copied from the cloud back to your office.
This way you can render over night in the cloud and on the next day you can start with the compositing in your office.

Multiple caching fileserver with metadata pre-sync

These settings are required for "B) Solution with metadata pre-sync"

Synced fileservers: Wait X sec. after submission before checking output folder.

Synced fileservers: Wait X sec. after client finished frame before checking for frame.

These two issues will be solved with these settings:

• Issue "Submission"
  Your rrSubmitter and rrClient are in Region-B.
  Your rrServer is in Region-A.
  If the rrSubmitter in Region-B creates the output folder or deletes old rendered files, then this is not immidiately applied in Region-A.
  The metadata might be synced in 10 seconds, it might take a minute.
  
  Without a delay, the rrServer would try to create the output folder directly and it creates a conflict if the rrSubmitter has already done it.
  Or if the rrSubmitter has deleted the old rendering, the rrServer still sees the old files and finishes the job.
  
  Same with rrServer and rrClient:
  If the rrServer creates the output folder direclty after submission and a clients gets the job, the rrClient might create the output folder as well.
  And again, you have a conflict.
  
  
• Issue "Rendering"
  If a client finishes some frames and the rrServer does not find them, then they are re-assigned to other clients.
  Therefore the rrServer has to wait until it can check the frames of that rrClient.
  
  

Multiple caching fileserver without metadata pre-sync

These setting is required if you have "A) Solution without metadata pre-sync".

If you use caching fileserver which copy the data after some time to the main fileserver, then the rrServer has to check the output folder on each fileserver to get a list of all rendered files.
Otherwise it would detect that there are missing frames and re-renders these frames.

Job Path: This is the begin of a path as specified in the output of a rrJob.
IMPORTANT: The Job Path will NOT be searched for files any more. 
If you want it to be searched, then set it as one of the caching servers as well (see linux example).

Note: If a job output folder does not start with any of the Job Paths you set in this config, then the rrServer uses that path without any change (default mode).

Cache only: 
A job is not set to finished as long as files are on Cache only fileservers. 
The files have to be on one of the other fileservers.

Example rrServer on Windows:

Example rrServer on Linux: