[John Heagy]

I’m trying to make a folder at the same depth for every Xsan volume and not have the user specify which volume. $i includes the full path all the way down to / including the volume name. The other path variables fall short. This limits any script based file commands using path variables to the same directory as the source file. If it’s complete in $i why omit it in the other variables?

We will solve it by calling a fully baked script but it seems like an odd omission if one wants to do things in other directories.

[John Heagy]

I don’t see any way to express a Xsan volume name as a variable. Am I missing something?

[John Heagy]

I think he understands that. He’s asking if the 3:2 cadence can be “reset” at cut points to avoid mixed cuts. The only way to do this is process every shot separately then put it back together. This avoids mixed cuts buy creates broken cadence potentially at every cut across the show.

The best way is to edit the 24p timeline on frames that don’t produce the mixed cuts.

[John Heagy]

[Oliver Peters] “While Randy U was certainly tight with the redevelopment of FCP into FCP X, there are plenty of competent engineers and designers on staff”

That is true and makes all the bizarre decisions even harder to understand. I think Steve “knighted” Randy King of Pro Apps and he became the “Emperor with no clothes” when it came to people standing up to challenge him.

To this day understanding how intelligent people, presumably in a meeting, could all be on board with killing FCP7 the same day as FCPX is released and have it not be able to open FCP7 projects.

[John Heagy]

Yes you do need a buttoned down workflow and a wide spread reliable shared storage environment. All of which we have. Our users understand the limits and advantages of ref mov. For us it’s mainly a time saver. We move 100s of TBs of files into Episode watch folders with custom metadata. Doing the same thing with self-contained movies would takes many minutes to copy each fie compared to a fraction of a seconds for a reference movies.

[John Heagy]

Thanks for the detailed response. Yes we have a Coraid system that uses an Oracle server running Solaris. The current 108 drive system can sustain 30 ProRes140 ingests and 50 growing file playbacks. That adds up to 1700MB/sec of “do or die” bandwidth. Unfortunately Coraid is on the ropes so we are looking for a way to duplicate the success we’ve had using another zfs implementation.

It’s really 3 approaches to choose from:
1) A turnkey system with limited options like with an Oracle “appliance” or ixsystems.
2) A wide range of supported hardware options using Nexenta based on illumos.
3) A compete roll your own with FreeNAS (BSD) or Solaris but without all the tools the Oracle appliance provides or any support from FreeNAS.

We did a lot of testing with the Coraid POC and found a few settings that unlocked performance so I do think we could build a nice FreeNAS system but it is daunting to work without a net support wise.

What kind of storage chassises are you using? The Coraid chassises are OEM’d SuperMicro https://www.supermicro.com/products/system/4U/6048/SSG-6048R-E1CR36N.cfm

We are very happy with the performance of these chassises, and I attribute that to the 5 internal SAS controllers Coraid decided to include. I believe the 8 drive per SAS card ratio reduces latency which is the main reason for dropped frames.

The drawback is the SAS cards are not active/active or even active/passive like ixssystems. Because of this we designed a 6 chassis system with 6 drive vdevs, one member from each chassis, meaning we can loose two chassises without loss of data.

How did you architect your FreeNAS system?

Did you consider Nexenta?

John

[John Heagy]

Hi Jon,

I would like to try FreeNAS. Why did you choose FreeNAS over the turnkey TrueNAS system? Did you have storage chassises already in hand?

John

[John Heagy]

Thanks Dougal!

I was hoping for a few “thumbs up” from others to help push it’s priority.

No one else would welcome easier TC searching?

John

[John Heagy]

Thanks Ivan,

How many hard drives in the pool? raidZ1 or raidZ2? Any flash or NVRAM for read or write caches?

[John Heagy]

Hi Bob,

Interesting comment. So the Solaris system was the NAS head? Was this straight up Oracle Solaris or Open Solaris aka illumos? What kind of problems?

What was the storage behind it? Oracle has an interesting zfs based storage system with killer reporting. The storage shelves are very reasonably priced, believe it or not, most likely due to the lack of RAID controllers as zfs handles all that.

Managing Solaris is a bit like managing OS X without a GUI. It’s just another Unix variant without any of Apple’s OS shenanigans.

John