[Warwick Teale]

Hello guys.. this is a very simple to fix…  Compressor.app V4.1.x DOES maintain some state information in the structures for EACH defined QMASTER cluster in these libraries. Simply and selectively DELETE the CONTENTs of these libraries (folders) and crack up the Compressor.app V4.1.x applications and these stale “Active jobs will be gone.

You just need to PURGE the stale STATE information for the stale batch job.. which has been maintained over a failure and not cleaned up.

This sometimes happens when a service node in the Qmaster cluster strays and goes away (network outage etc ??)

NOTE: myth busting time: there is NO NEED to delete Compressor.parms, or reinstall the compressor.app app nor use any third party removal apps! These errors are usually the result of stale STATE information.

Although this MIGHT fix your these issues, it removed any of you customised Compressor.app SETTINGS, templates, droplets,locations, and HISTORY the latter being highly useful for RESTARTS and REDOS… and is highly disruptive and usually unnecessary during a client job!

So.. Try this: on each host that is part of the cluster or where this stale “Active” UI shows the dud jobs:

1) in Compressor.app v4.1.x preferences/my computer/”allow other computers..” = OFF (stops qmaster and releases shared storage structures)
2) (OPTIONALLY) quit the Compressor.app V4.1.x UI.
3) use the OSX Finder to navigate to your home directory’s ~/Library/Application\ Support/Compressor/Storage or +shift+G and paste in the string path above.(use backslash for blank spaces in file path!)
you will see folders) that are names if hexadecimal strings like xxxxxxxx-yyyyyyyy – one for each defined cluster.. You may only have one. Inside each cluster you will see folder structures with names like (Jobs and Shared etc)
now INSIDe the JOBS folder will be some structure and some HIDDEN objects such as this little bugger .qmaster.plist that is causing the issue. (Just delete it all.
as an example…. you can see is as ls -a1r ~/Library/Application\ Support/Compressor/Storage/94941F8A-4FFBF45C/jobs
3) now DELETE the the cluster storages xxxxxxxx-yyyyyyyy or their contents of these Storage folders.
use the unix rm command to do this rather than the Finder Trash because an empty trash action may fail whilst serialising on these folder names.
4) in terminal.app and rm -rf ~/Library/Application\ Support/Compressor/Storage/xxxxxxxx-yyyyyyyy (deletes ALL the cluster info) – (use backslash for blank spaces in file path!)
5) next, Finder navigate to /users/shared/Library/Application Support/Compressor/Storage/xxxxxxxx-yyyyyyyy
6) Delete ALL the individual clusters storage .. there may be more than one.
as above , use the unix rm command to do this rather than the Finder Trash because an empty trash action may fail whilst serialising on these folder name
7) in terminal.app shell: rm -rf /users/shared/Library/Application \Support/Compressor/Storage/xxxxxxxx-yyyyyyyy
8) re-start Compressor.app V4.1.x
9) compressor.app v4.1.x preferences/my computer/”allow other computers..” = ON (starts qmaster and initialises the shared storage structures)
10) check your clusters in compressor.app v4.1.x preferences/ “shared computers”

Make sure you repeat this on all SERVICE nodes in the affected cluster.

Note we have 2-3 active any any time depending on the length and complexity of the transcode work.

should be right

HTH

The stale ACTIVE jobs should not show.

Also lots of good inform on compressor V4.1 on the  forums too.

Warwick
Hong Kong

[Warwick Teale]

Hi Tom ,regarding the error you see, (error: mkdir failed for path = [/Users/Shared/Library/Application Support/Compressor/Storage/D8220947-D02E6012/shared), just a couple of things you look at:

1. if you are runoff OS server, make confirm the path is shared with access to you r/w (.. it usually is
2. is this cluster storage has been removed previously, itMAY still have similar directory in the .Trash directory. ou may need to remove both of these .

If 2. is true, then

a. stop sharing (don’t share computer in Compressor.app 4.1 prefs
b. delete the all the structures (folders) in /Users/Shared/Library/Application Support/Compressor/Storage / directory
c. attempt to empty the trash on your files systems. (cmd+shift+delete)
d. OPTONALLY: if the Trash won’t empty it may be because the serialisation needed to delete the objects in B. is not being honoured. Try to “rm -rf ~/.Trash” contents. If that fails you may need t restart th MAC to resolve the issue d. . The attempt steps c. and d. again

A few people have had this issue.

It may help you
warwick
Hong Kong

[Warwick Teale]

Hi Mark Thanks for the tip regarding deleting the FCPX V10.0 app from /Applications and going to apps store/purchases to reinstall to get the update.., that worked like charm. I had the issue were FCPX 10.0 would not update to V10.1 although/however the apps storer updated Motion5 and Compressor.

W
Hong Kong

[Warwick Teale]

Thank Bob! good advice. 4 x WD RE4 WD2003FYYS’s it is. I shall order now.

Thanks again for your reality check 🙂

BTW were you on the MATROX booth in LVCC south at NAB? if not, that lad sure looked like you from your picture! lol

w
HK

[Warwick Teale]

another excellent alternative I\’ve used for a while with catdv is a combination of bru-pe+HP 1840 or HP 1760 lto ultrium sas tape drives+atto h380 hba

Yep this assumes a desktop machine with pcie slots! much cheaper that the cache-a solution btw. Lots of change left over.

Replace mt lto4 ultriums with newer HP lto5 ultrium and the plausible but yet unproven LTFS and maybe ditch the software app.

Only workflow issue is I have no tight integration between catdv and bru-pe yet. So I do this manually

I also have added user folds in the catdv metadata for tying my stuff together

Works ok

Btw I have 24TB of disk and I\’ll tellbyou that the lto4 tape archive has saved me many times when disk arrays have failed.

Hth

[Warwick Teale]

Thanks, yes Rolf @ Squarebox contacted me. Yes this is ordered through a reseller. I have initiated the purchase this particular licence.

Thanks for your help.

w

[Warwick Teale]

HI Rob, just a heads up. I am too having tremendous frustration with Motion4 (V4.01) on this 2009 MA PRO w/16GB Radeon 4870 and fast disk array where I have placed all motion caching file (this is not the issue btw) :).

I see one single vcore at 100% a lot (apparently sending work to the RAdeon 4870!). the other 15 vcores sit idle. I too have 10-15 secs trying to scrub or worse 20 secs while waiting for the HUD to change when I click on a behaviour and also some other elements. (i’m rather behond on hte project at present).

I have several nested layers with a few behaviours in one of these 15 secs projects. All have the same issue.

I employed all Mark Spencer’s (and others 10 tips+ to make motion more efficient – but have no real effects over all). (soloing layers and objects etc etc)

So HEADS UP that via trial and error: You might want to SELECTIVELY DISABLE THE BEHAVIOURS on the layers for the layers that you have specifically disabled! This seems to reinstate responsiveness of Motion.

Turn them back on when you are ready to bake.

So far seem to be working ok.

Now to catch up 24 hours of work.

Let me know if this works for you

[Warwick Teale]

HI Jason, I’d agree that 300+MB/sec is slow but my observations are not from NOT using the AJA system test tool.app just some basic non capture workflows (finder and shell commands for moving stuff round and FCP work, compressor.app and motion.app etc).

I also only have 2 groups x (4 x 1TB WDC1002’s , 4 x WDC5000’s 500GB ) each R380 group, not 8 x DDMs.

R380 HBA on slot4 (x4 mode)

Using the AJA System Test tool for DVCPROHD1080/50 4GB object on BOTH of these 4 by RAID0 groups I get:

(a) group01 RAID0 striped over 4 x 1TB of 7200RPM WDC1000-YS with results when
• disable file system cache is ON (ticked): WRITES=390MB/sec and READS=430MB/sec

(b) group02 RAID0 striped over 4 x 500GB of 7200RPM WDC5000YS with results when
•  disable file system cache is ON (ticked): WRITES=279MB/sec and READS=261MB/sec

Shoudl I get time to tie all these WDC DDMs and configure a new 8x GROUP into an 8 x Raid0 on the PROAVIO E8MS JBOD enclosure after i arcive the objects, I think I will se at least a 50% inprovement in the realm of 500MB/sec as other are seeing.

However for my workflows this is fine.

As for the original post my ATTO TEch R380 SAS HBA is on SLOT4 as an x4 lane. works great!

(i also have an LSI SAS 3442E-R in slot 3 as an x4 lane in MACPRO because I have some SAS ULTRIUM LSO tape drives attached from time to time).

hth

w

[Warwick Teale]

HI Jason, just saw your post and wondered whath throughput you are getting for:
• captures &
• objects copies via the OSx finder or for example shell’s “cp”.

I believe I have a similar setup to youself as a pre 2008 MACPRO 8 x core (8GB ram)..

I also have an ATTO express R380 HBA in slot4 (x 8 and x4 [testing]) with both multi lane minis sas cables SFF-8088’s into a PROAVIO EB8MS JBOD enclosure (4 x 1TB WDCS 7200rpms and 4 x 500GB 7200rpm WDC DDMs) (4+2=6TB) . I configures 2 x RAID) stripes with ATTO config tool as:
• 1MB stripe
• 4K sector
• x4 read ahead buffers (for when I pull stuff around).

Using expansionslot.app and the ATTOTECH expressas R380, I have conducted testing using controls of:
•  SLOT4 as (8x) and
• SLOT 4 as (4x)

with DVCPROHD 1080p (110Mbs) material to see how long it takes a few hours (80-100GB) of this stuff to be around and around these two external RAID0’s in the PROAVIO EB8MS.

I’m curious to know what effective data rates because I have tried various objects movements between these two RAID devices?

I seems to be maxing out at around only 305MB/sec….. which my all accounts on many forums is not that fast for this setup.

I also have an LSI SAS3442E-R I use for ultrium LTO4 Ultrium SAS tape drives (works great BTW) with HP and QUANTUM on OSX. (slot 2 or slot 3 as x8 or x4 respectively)

Look forward to your reply.

thanks

w
Hong Kong