Qmaster/compressor issues

I set up a Q master cluster for compressor, but every compression immediately fails. What might I be doing wrong?

how many computers, what version of FCP, ... need much more information.. are you trying to export out of Final cut pro?

Similar Messages

ECC / Parity errors but 'only' when running qmaster / Compressor ?!!

I have a Mac Pro that came with 4GB of (Apple) Ram. I recently installed 4 addiitonal (to fill both risers to 8GB total), of branded, supposedly apple-certified Ram from one of the major vendors we all have come to love and trust.
Anyway, all seems fine - but now I am running QMASTER/Compressor which in fact wires (or at least pages) so much ram as to make good use of my 8: I have only 80MB free- but hey that's good, the memory is being used.
Here's the odd thing, when doing compressor / qmaster tasks (this is both the cluster server and a service node) I now get ECC Parity errors:
- Of my 8 chips, 3 have trown ECC Parity errors (all corrected), ranging from 1 on one chip to now 35 on another. This is in a 24h period
- THis is on the APPLE ORIGINAL RAM - none of the 'new' chips have any errors ?!!!
- Errors show up in 'about system' as well as the console log- but I ran EXTENDED Hardware test twice - and both times the memory (whole system) comes up clean!
- ECC errors don't show up - as far as I can tell, if I don't 'extremely' tax the system - i.e. running email, Safari, iTunes and, maybe WoW keep it 'clean'! but also leave between 3-6GB memory 'free'
- The errors occur on 2 seperate risers
So, here's my dilemman - I can replicate this everytime i use compressor/qmaster. However, I never (yet) had a related crash or issues.
It started with 'just' one chip, now it's 3 - again, all the original apple ram.
My memory is installed correctly (i.e. both apple pairs together and both OWC together).
I would LOVE to get some suggestions. Logic would say this is related to the new memory BUT I also never ran a compressor/qmaster renderfarm before now. I already opened a case with Apple and they are sending me one new chip (I opened at the time only one chip was having the errors).
So - normal? Low enough at high volume I shouldn't worry? Routine with compressor and qmaster Or time to panic?
THANKS!
Dan

This is very interesting. We, too, recently experienced a hard RAM Parity Panic/Crash on a Qmaster/Compressor encoder node. About two weeks earlier, we experienced a RAM Parity Panic/Crash on a system that has nothing to do with Qmaster.
Both machines are dual G5 Xserves with factory-installed RAM. The encoder had six 512MB sticks while the non-encoder server had 2 1MB sticks.
The encoder died, but when rebooted kept working by ignoring the bad RAM and working off only 2GB of the good RAM (must be in pairs).
Apple replaced both sticks under warranty.
As for Qmaster/Compressor causing the errors, I have no insight. If that were the case, I assume we'd see this on most if not all our encoders and cluster controllers. And that hasn't happened.
I'd say, you got bad RAM and it needs to be fixed.
The panic.log should give you insight into which dimm slot threw the error.
Good luck!

Compressor issue, keep getting Failed: QuickTime error.-50. Anyone knows why?

Compressor issue, keep getting Failed: QuickTime error.-50. Anyone knows why?

I have this problem too, I sent a 1hr long video to Compressor from FCP and trying to save it in H.264
If works for a few hours and then says it failed.
Any help would be much appreciated!

Qmaster Compressor and Shared Storage

I've got Qmaster working on my compressor encodes and everything is working fine. What I want to know is whether there is a way to cut out the extra copy of data for the encode. All of my machines in the cluster have the same shared storage mounted on the desktop, but still my jobs get copied first and then encoded.
Is there anyway to configure qmaster, compressor and the cluster machines so they all use their local path for source files? And avoid the extra copy of the source files to the controller system?
Thanks in advance.

You want to make sure in Compressor, under Preferences, the Cluster Options field says Never Copy Source to Cluster. Assuming you have all of the correct permissions set and have the Shared Storage configured properly then it should work. "Should" being the key word.

How does Qmaster/Compressor distribute tasks?

I've been reading up on all of this and I don't understand how Qmaster and Compressor divy up tasks to the connected CPUs.
Lets say there's a sequence in FCP that needs to be exported to DVD. Does Qmaster/Compressor send bits of the audio and video to each CPU in the cluster or does each CPU get one task, like audio gets sent to one cpu, video gets sent to another cpu and any left over cpu's are left idle?
I guess I'm asking if Qmaster/Compressor divides up the work among all the CPUs for FCP exports/transcodes or does it just assign one task (audio or video) to a dedicated cpu in the cluster. I can't figure it out.
Thanks!
Too many to list Mac OS X (10.4.8)

If you use ref movies tho you can add machines that don't have FCP installed to your cluster (this does not violate the license terms either.)
The only thing w/ use ref movies is that all nodes need to be able to access the referenced footage at the same url, ie:
afp://myserver/sharedfootage
as long as sharedfootage is up, you are all set...
Good to know about the timeline tho - can you still segment jobs when coming off the timeline?

Disabled security for Qmaster, Compressor won't submit batch

I have read anything I could possibly get a hold of to fix the "unable to connect to background process" problem. I have tried them all except for reinstalling because after I installed a software update for compressor, it actually worked! But when I restarted guess what happened..the same darn problem. I tried manually turning on my Qmaster but it was already on. So here's where I believe things went wrong: When I originally installed FCP5 on my mac this message came up when I restarted:
An item in the Startup Items folder ("/Library/StartupItems/Qmaster") does not have the proper security settings.
It gave me the choice of fixing it or disabling it, well guess what I did...I disabled it like a moron. Does anyone have any suggestions what to do?

I had the same "moronic" issue and tried to reinstall, with no avail. What I found is that you must make sure to trash Qmaster from the "library/startupitems" folder before reinstalling. Then it will give you the same prompt to "fix" the security settings at startup.

FCS3 - Qmaster / Compressor trouble still exist. Same work around works...

Howdy...
With FCS2, I was not able to get a quick cluster working on my octo Mac Pro. It basically would not show up in Compressor no matter what I did...
Finally, I stumbled upon the network being the issue. If I pulled my ethernet cable or deactivated the ethernet port, the quick cluster (set to 4 instances) would show up, and I could submit jobs to it. Everything worked fine. Once I had submitted the job, I could reactive the network (or plug the cable back in) and the cluster would stick around until the end of that job... It was a pain, but it worked as a work-around...
Anyways, I've just installed the new FCS (3???), and the same problem exists, so either it's an issue with my installation (but it's not bad enough for me to do a complete format/reinstall) or it's an issue that wasn't addressed by this update.
Thought you'd like to know.
Best,
Ben

Same problem here. Mac Pro 8-core. Quickclusters were working but then disappeared about a week ago. Tried everything. Installed the FCS 3 upgrade yesterday but that didn't help. I've reset the Qmaster services, re-configured, restarted, but still no Quickcluster shows up as an option when I submit a batch to Compressor. I had this working for the past year and then it just stopped. I tried disconnecting from the network but that didn't help. What's the solution?

FCS Compressor issues - Quartz Extreme not loading on headless XServe

Hi,
I am running FCS on a OS X Leopard Server Intel XServe...installed a Quartz Extreme GPU (an old stock nVidia 7300 taken from a Mac Pro) to meet the specs. Everything worked fine. However I want to run the system headless - it is an XServe in a rack! Only problem is that without a monitor attached to the GPU, the XServe fails to acknowledge the existence of the GPU. Quartz Extreme does not load. Compressor will not launch if opened (explaining that I have no GPU!) and the automatic encodes from FCS come out as 0Mbyte .mov files. I admit I already knew about the issue of Final Cut Studio apps not running on a headless machine, even with a GPU installed. I just assumed Apple must have fixed this one for FCS because it is server software....I am guessing plenty of people must actually run FCS on headless machines....what's the trick to getting encoding to work?
Thanks!
Daniel

Depending on your xServe, you can just stick the dongle in the back and it should activate Quartz Extreme. You shouldn't have to install another graphics card. If you have an older xserve, you may have to hook up a monitor to the dongle.

Compressor Issue

Lately I've been experiencing issues while trying to export from compressor. The issues lies in the fact that when the process begins to export a file it severely affects my whole system for example it will not tell me an estimated length of export it now says "unkown" and then begins to freeze up and crash but when I close the application it acts as though its still open affecting all other applications. I have to force quit and restart my whole system to get it functioning normally. Any suggestions would be helpful. Thanks.

First thing to try is [Compressor Repair|http://digitalrebellion.net/compressor_repair.htm].
Also check out the other Final Cut Studio tools on the site, it is an invaluable resource for editors and the author of those applications is a regular on these forums.

Trouble with Qmaster, Compressor

Hello all!
I am trying to compress a 48 Minute long HDV sequence to HD-DVD using the H.264 60 Min encode settings, and I am having problems.
I have read how to setup Qmaster to run multiple cores, so I have made a 4 core quick cluster on Qmaster with the following settings
"Quick Cluster with Services" - Checked
Under the services box, I have "Share" checked for Compressor
I have 4 instances selected (out of 16 possible)
"Include unmanaged services from other computers" - Checked.
All other options unchecked.
On Advanced menu, only "Log service activity to file" was checked.
What happens is that when I submit the job in compressor (I select the 4 cluster when I submit), I see all the instances in Activity monitor, but one of them is red and non-responding. I can't get all the instances to do work at the same time, as the cpu utilization is only indicating one instance doing work at a time. I thought they were all supposed to be utilized.
In the end, I waited 36 hours, and the encode still wasn't finished.
Any ideas what could be going on? Am I set up correctly? Let me know any thoughts!
Thanks,
Ken

Hi Ken here's some ideas. I use compressor/Qmaster all the time. I too have a brand new Nehalem MACPRO 2.93 @ 12GB.
Firstly for distribution, *compressor is your best friend*... see why...
Some diagnostics tips first:
• go check out /var/logs or the utilities/console.app (leave the late in the dock). Look under the various directories for QMASTER. There is usually a wealth of information in there.
• COMPRESSOR.app/prefs: not always but worth it to (a) MOUNT the shared cluster storage you used in /system prefs/qmaster/advanced: shared cluster storage
• COMPRESSOR.app/: look at HISTORY
• Batch MOnitor.app : check on the "i" status for each node running a segment in the multi pass segmented transcode (segment=TICKED in inspector video for compressor). Look at the "status" and "logs" in the pane.
• turn off ALL network devices (airport, EN & FW) until you get it working QMASTER seems to go looking everywhere for other nodes.. so dont let it.
• COMPRESSOR.app/prefs: set "NEVER COPY SOURCE TO CLUSTER" - we'll work about his when yu start to use other nodes outside your MACPRO.
• COMPRESSOR.app/reset background processes - to start
OK when you have all tjis done... let look at some ways to commence to fire up Compressor using Qmaster cluster.
• assume QUICKCLUSTER ste up: click "managed services=ticke" if you like
• *systems prefs/qmaster:* set SERVICES: for number of instances I am using 14! (*yes 14 instances*) on this MACPRO and it rides all the 16 cores and I can edit fine on FCP at the same time..PLenty of head room in these new MACPROS 2009 models (+doubters go buy a dell p.o.s+ ). *CAVEAT= YOU HAVE ENOUGH RAM* - not sure if 6GB will cut it so trial and error.. works great for me thou and I have 8GB not used from 12GB!. NOte in the non Nehalem MACPRO era the R.O.T. was one instance of compressor per core... however since I have bene usingthis baby on the last few days I have cranked it to 14 instances and it runs fine and all cpres are nearly 100% usage with a very responsive machine. THe transcodes are at least 3-4 times faster than my 8*core 2007 MACPRO!.
• *systems prefs/qmaster:* / advanced: tick the port ranges =on.
OK .. now you're ready to rock!
1) *systems prefs/qmaster:*: setup: clock OPTION=CLICK over the start button to RESET any transcodes that were going.
2) Set up a job in compressor.app making sure "job segmenting" is ticked in the ENCODER inspector.
3) submit and select your cluster in the list box.
4) sit back while your new MCPRO 2009 earns its money! - watch all the cores light up GREEN in the activity monitor!
5) watch teh console.app "all messages for errors" form QMaster.
OK.. this is only a summary.
POst your results of PM me if you need help.
I love compressor and Qm especially on this NEW MAC PRO.. it just smokes anything else I've used for transcoding.
Oh: *and one other tip for your workflow* which you can take or leave regarding HDV. Yep its not on may favourite codec list but here's some good tips for getting your compsition ready for distribution as you are trying to do from when I owned a SONY-Z1P HDV some years back.
This will cut 10's of hours of what you are seeing. (well maybe not that magnitude).
With the edit prepared and not rendered:
• take ALL the footage fomr the HDV that you ingested form the tape in the CAPTURE folder and use COMPRESSOR.app and transcode it to PRORES422 only. (Same resolution etc etc).... Simply try this out please. USe "copy and BLOCK SELECT + paste all the codec+targets information in the COMPRESSOR window.. its simple... submit it to your QMASTER and come back in say 10 minutes .. it shoud all be done. (what you have done is to make a mezzanine instance of LGOP HDV footage into iFRAME PRORES422 - no quality has been lost nor gained.
• in your FCP project DISCONNECT all the media from your Sequences and then RECONNECT the media to the same file names that are the new PRORES 422 - (conforming) in FCP.
• now your edit sequqnces are pointing to the PRORES versions of your clips. edit at will here and watch the play head and edits slice through prores422 like the proverbial +hot knife through butter+ (assume you have a fast disk system underneath.. my MACBOOKPRO does this in its sleep).
*when your ready to make a distribution*, try this from within FCP timeline/sequence:
• export QUICKTIME MOVE" as a reference movie (untick "self contained") and call the movie client_ref.mov (etc) . *This step is very important*. And +DONT EXPORT TO COMPRESSOR+. this is nuts at the moment. wait for FCS V3 soon I hope.
• last step: make this distribution using COMPRESSOR by using the reference clip client_ref.mov as input to compressor.
And as they say .. thats it!
Ph one more cool thing.. while it is transcoding and all your cores are ringing out at 99% each, try setting up a COMPRESSOR DROPLET. Set this on your desktop and drag and drop client_ref.mov onto it.. and watch compressor and master do their work.
Yes.. compressor is your best mate when you set him up properly.
HTH
w
sorry for the typos, I'm in a hurry...

Qmaster/compressor- only 2 of 8 render farm nodes service compressor batch?

I have 8 nodes in my farm but only two are servicing Compressor jobs.
In the QM Administrator console these two nodes display a line item for both "Compressor" and "Rendering" in the "ServiceType" column, the other nodes only show Rendering so I can understand why these two are the only nodes that are handling the Compressor batch.
What I don't understand is what I need to do to configure the other nodes in the farm to also display a Compressor "service type", I've checked and rechecked the QMaster settings on all the machines and there is no apparent difference in the configurations b/t the two nodes that work for compressor and the others that do not (as viewed in the QM System Prefs panel).
FYI, there are no other batches running, ALL nodes correctly service other render jobs, such as Maya and After Effects, and there are other Compressor batches "waiting" so there is enough work for the other nodes to complete.
Any ideas are appreciated.
Thanks,
Brian

I answered this one myself. From Sys Pref->QMaster on each node:
1. Stop Service
2. Option-click "Start Service" button to change button to "Reset Service", then to Reset Service (note: resetting the service will remove the node from a Cluster if it belonged to one).
3. Start Service
4. In QM Administrator add the node back to your cluster.
*I am using managed services only and each node has "shared/managed" enabled for BOTH "rendering" and "compressor".

After install 10.5.7 update QMaster/Compressor hangs/not responding??

I find it funny that the 10.5.7 update was supposed to "improve OS stability". Well it looks like all it has done is made my Mac a worthless pile of pooh.
I have an older MacPro (2007) with two dual core processors (4 CPUs). I setup QMaster as QuickCluster with services (this is the only way Compressor will see it when I submit a batch). I selected 2 CPU instances. Hit "Submit" in Compressor and open the batch window, I see it "try" to parse out the video segments to 2 CPU instances, but get Audio successful and Video "Unknown" with a Service controller "error" and a progress bar for the main batch at 50% with time duration increasing (never decreases).
So I bring up Activity Monitor and find that coreaudio is "not responding" compressor is "not responding" and final cut pro is "not responding" ... oh joy!!
All this "used" to work before the 10.5.7 update, so what gives???
How to reproduce on my hardware:
1. Open FCP
2. Create a Sequence
3. Send Sequence to Compressor
4. In Compress I have a predefined H.264 setting and destination with "Job segmenting" checked, hit submit
5. Open batch window, look at 50% mark with errors mentioned above
NOT very happy.
Rob

Compressor repair? Can you give me more details on what this is?
I had FCP to Compressor using QMaster (2 CPU instances) working in 10.5.6. It created some bizarre project name under FCP (i.e. 13DAA64E-8A34-4476-A12B-86) but would work. With 10.5.7 the same steps just hang/fail -- fail as in they Submitted batch never completes (I can leave it for 10 hours on a H.264 small duration clip of 30 seconds).
So now (with 10.5.7) the only way I can get Compressor and Qmaster to work is if I export sequence to QuickTime movie, close FCP, open Compressor create a new batch using the exported QT movie, set my settings/destination then submit. Which defeats the purpose of using a 2 Instance QMaster in the first place since I've had to introduce a Export to Quicktime in FCP. So essentially any benefit using 2 CPUs has been negated by the extra export step.
Sorry for the frustration/tone, but if I wanted this much hassle with Mac software/OS, I would have just stayed exclusively on a Windows based platform -- as of right now I see very little difference between the two companies, both are not responsive to correct obvious problems and both create more work than solutions. I gotta wonder why I even selected to use a Mac and why I successfully converted at 11 friends over from Windows to Mac - SSDD.
Frustrated.
Rob

Qmaster/xsan issue?

I posted this over in the qmaster boards and I got some quick replies that brought up xsan so i figured I'll post it over here as well:
I used to notice this back in the day with 24bit audio exports... it's been since fixed, although now it's returned in a far more brutal way
I have 3 machines on a qmaster cluster... an xserve, and two g5 towers. The xserve works as the controller. I recently submitted a whole batch, including files that are either animation or 8bit uncompressed, with either 16 or 24 bit audio. I have each clip compressing down to a mix of h264, sorenson and mpeg4.
After 2 days of exporting, i opened up the output folder and checked the files and notice of 288 files, 65 of them return a "-2048" quicktime error and do not open. The funny thing, its completely random which files get it (not related to certain export codecs or input codecs or anyting of the sort)
So either way, I got no idea what's going on. I'm using Qmaster 2.0.1 across the board, and the only thing I could possibly think of causing this is that I'm clustering off a xsan volume (1.3).
Any ideas would be greatly appriciated.

No, FCP is not installed on the G5-1.8 but compressor is along with Q-Master. Should FCP be installed on the 1.8?
Sky......

Qmaster compressor and wmv

Hello,
I have constructed a qmaster cluster with two 8 core mac pros.
The cluster was manually configured using Qadministrator. One Mac Pro is the cluster controller with services and one is just services for the node.
When I submit batches to the cluster using the mpeg 2 preset in compressor, the cluster uses all 16 processors and it goes really fast. When I submit using a custom preset that exports to WMV using the regular Flip4Mac (not episode) it compressor only uses the processors on the cluster controller. My rendering time has to be as short as possible and I don't understand why a cluster which appears to be properly configured given that Mpeg takes advantage of all 16 won't export wmv the same way.
I have read that the codec matters in how the jobs are sent to the cluster.
Also in settings MPeg can "create job segments" and wmv does not let you check that box.
thanks
ben

This sounds like a good question for Craig at the Flip4Mac forum

QMaster & Compressor - Multi machine control

I don't necessarily want to create a cluster. What I'm trying to do is run all of the compression machines on a network (running Compressor) from one application/window. Is there a way I can select the file to compress, select the machine I want to use to compress it, then select a compression profile for the machines to use? Maybe something similar to the batch processor with a few more options/network support?
I'd like to have a set of common compression profiles that all the machines can use, too. Instead of creating new/changing settings on each machine.
Any ideas on setting machines up like this? The problem, is that I need to run files through the compressor network system from different workstations. Just trying to find a simpler method than using remote desktop.
Thanks!

ou only need compressor on the machine that submits. Put QMaster on the others, and configure them as clusters of 1 machine.
You can then select which 'cluster' you want to submit to in the Compressor app you use to submit.

Qmaster/compressor issues

Similar Messages

Maybe you are looking for