Problems with Compressor 4.0.7

Try running this.

As to upgrading to 4.1 and 10.9, wait until you're not in the midst of an important project. Back up everything and read up on the new app.

In fact, if you have a spare drive available, create a bootable disk; install Mavericks and upgrade both FCPX and Compressor on it. Play with it for a while. Judging from the threads here and elsewhere, you'll either find it a notable improvement or a disaster. (I've found both apps to be faster and much more solid.) The people that seem to report the best outcomes are those who do clean installs.

Good luck.

Russ

Hi Mr J, there's many post on fixing this in these forums for Compressor.app V4.0 and earlier. Its simply the STATE information for QMASTER is hosed. You just need to purge it. By all means use a 3rd party app if it works for you .Be aware you might lose older job settings.

If you have not yet tasted success of tried Russ's advice...

Rather than type all this info out again.. look at posts and many others like it.

https://discussions.apple.com/thread/3197005?answerId=21309544022#21309544022

it merely :

• stops QMAster (do on all hosts)
• rips out the state information
• make sure the share storage is active
• and fires up sharing/Qmaster
• redfines the cluster
• sub the job

Post your results for others to see.

W

Hi Jon, a "cluster" in this context of Compressor.app is to logically group a set of resources together to be used to TRANSCODE (or other processes such as RENDERING etc) to IMPROVE EFFICIENCY (mostly reduce elapsed/service time) by utlising distributed CONCURRENT processing (not strictly parallel BTW) and file sharing.

Segmenting (AKA "Job Segmenting" ) in this context of Compressor.app is a method of intelligently sectioning pieces of the SOURCE material to so that they can be distributed for concurrent processing over many INSTANCES on many Hosts (servers | service nodes) and then reassembled into the final distribution (encoded object.. an .mv4 object for example). Put crudely: "cutting the job up and spraying it over as many instances (on CPU cores or latterly GPUs) on as many servers (hosts) as possible to REDUCE the transcode (or other process) time.

BTW an instance in this description means a subtask of the transcoder. For Apple Qmaster subcomponent it is has the name of compressord. (You can monitor these in the Activity Monitor.app, Share Monitor.app or various Unix commands)

In addition, the facility used in MACOSX called GrandCentralDispatch with multithreading CPUs does a farely good job of utilsing the CORES and instances in a single host even without a formally defined cluster! Often dispatching a ttranscode on a single host (not distributed) will resultin a reduced elapsed time.

Apple Qadministrator.app is a legacy utility (pre Dec 2013 and NOT available in Compressor V4.1) to administer (create/ manage) Clusters based on Apple Qmaster. These formerly administered were loosely referred to as "Managed Clusters". (FWIW the simple cluster utilising many cores / instances of compressord on a single host could also be enabled using the Quickcluster facility which most people intially used. We found it quite unstable and unusable with many service nodes in most cases 😠. However others had great joy using it 😎).

MacOSX utilises the Apple Qmaster component to perform this.

Getting Started: A great place to start / investigate is in older Apple  Qmaster Manuals here athttp://support.apple.com/manuals/#professionalsoftware. To access the legacy manuals and documents (pre FCPX 10.1 / Compressor V4.1 etc Dec 2013) , make sure you click "Load More Results" or use the "search" field.

There's some very good reading material there while you lounge on the beach with your ipad!

Lastly, as Russ correctly points out that when your are ready to upgrate to OSX 10.9.x and compressor V4.1, the process to implement / deploy transcode clustering on a cluster has been wonderfully simplied!.

Post your results for others to see.

warwick

Hong Kong

Hi Jon, unless you need to use clustering, I'd promote just submitting the job to your current machine (on Compressor.app V4.0 and previous , this will be to the node "This Computer" until you are comfortable with concept of clustering and then distributed clustering.

My purpose in directing you to that post was to guide you through well worn path of steps to clean up and enable Qmaster and to alert you to the state information structures that ought to be purged (cleaned up) to initialise a fresh start of compressor.app V4.0 and previous. 🙂

Warwick

HI Jon and other reader, this is a reponse in reference to Compressor.app V4.0 and V3.x prior to December 2013 only NOT for compressorv 4.1.x (postr Dec 2013)... as you requested. My response should be viewed in context with Jons original post and an old post at link https://discussions.apple.com/thread/3197005?answerId=21309544022#21309544022 at 21 Jan 2012.

If you are NOT using formal QMASTER clustering then you may cease reading at this point. 😎

jfriedman93n3h wrote:

Warwick,

Thanks for all the info. since I do not need to use clusters can you clarify and break down exactly what I need to do in steps 11-13? Sorry for the inconvenience but I still am a little confused. Also for a Mac w/16GB RAM I would use 8 instances? Thanks again for your help.

10. restart/Launch Compressor.app V4

11. use the Compressor.app | "Compressor" application menu | Apple Qmaster | Share this Computer & go in and resetup your cluster. Under options setthe number of compressor instances to this rule of thumb compressor instances = rounddown(Amount of RAM in GB / 2) ... this is used for SEGMENTED TRANSCODING (use segmenting). If you dont need Rendering (Shake Maya, Nuke) then leave Rendering unchecked. YOu can use a Quickcluster howevere I've found for me and others that a managed cluster (use Apple Qadministrator.app) seems peculiarly more reliable... beats me.

12 now take a breath for 30 secs to a minute.... wait for Qmaster to get up an running.... then.

13. in Compressor.app, Mount the cluster storage .. (+shIft+M). Now this is a good test. IF THE CLUSTER NAME COMES UP IN THE LIST it means THAT THE CLUSTER infrastucture is working.... no please be patient here.. this could take a minute or two.. depending on your system.. and also do

14. n Lion OSX 10.7.2, use your pen/track pad (or mouse thing) and hover over the COMPRESSOR.app icon in the dock. Then right gear down / control+click/ two fingers/ right mouse button on the COMPRESSOR.icon and see ifthere are any ACTIVE SERVICES there for a cluster. You should see the number of compressor services to be the number of instances you specified earlier in step 11.

15. now... add a movie and try to submit the batch to the cluster. Hopefully the submit button will not be greyed out. (DONT SUBMIT FROM FCPX just yet..... just add a job a movie in Compressor to see if it works).

Step 11: is a reference to you customised specification for enabling additional INSTANCES of compressord (or a renderer engine). It simply kicks of another or a set of substasks for the cluster. Unfortunately a large specification (example would be 5 instances for a 16 core mac pro) may drown the mac (and other macs) and be of no efficiency benefit. This is especially bad for a Quickcluster and a render of Motion project for example die to the MEMORY utilisation for the MOTION project. THis differs again for a straight forward TRANSCODE operation.

• for your purposes of getting started and making you cluster stable, I would propose you DONT ENABLE these settings.
• it can be very very touchy. HEnce you would implement several clusters if you ha dthe resources to do so tha twould be used based on teh TYPE of workflow you had.
• it may transpire that TRanscoding distributed CLUSTERING is of little benefit..

Step 12: allow some time for QMaster to become enabled (startup) and settle.

Step 13: Qmaster clusters require some simple repository for sharing elements and targets BETWEEN hosts (service nodes on other systems - macs). In additional it will use this STORAGE on a single host as well when no other service nodes are defined the cluster. This step is a method of VERIFYing that the storage is available to the cluster PRIOR to submitting any jobs. Frankly it was problematic and hit and miss. THE verbiage in the RED says it all.

Step 14: was an undocumented way to determine WHAT Qmaster compressord subtasks (instanced you may have defined in step 11) are actually started. These would be manually added in a managed cluster using QAdministrator.app

Jon if you have managed to implement Compressor.app V4.1.1 since then these procedure are now outdated.

HTH

Post your results for others to see

Warwick

Hong Kong