[Steve Modica]

[Kevin Christopher] “I did a preliminary test with that, but It does not recover the file names. 8 TB of QT0001.mov. Blehh!!”

I think that’s the best you’re going to get. If the root directories are gone, the inodes no longer have names.

Files don’t really have names. They have inode numbers. Directories are lists of names that point to inode numbers. A file can have many names. It must have at least one name. You can create several hard links to the same inode (in fact, many unix commands like cp and rm are the same binary.. just linked with a different name). If you delete one of those names, the file remains (so long as there is at least one name).

You have a ton of files that no longer have names. So any utility will recover them as “generic nameX” because that’s all it can do. They can look at headers and magic numbers to try and guess the type of file.

Steve

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

The raid itself was not destroyed. You don’t have to rebuild that. This was a filesystem reformat (disk utility). So the RAID is happy (or should be except that he power cycled it without warning).

So the main issue is superblocks.
In the olden days, we could find a “good” superblock and run mkfs is a recovery mode pointing to that block. It would layout the filesystem again without touching the data itself. Then you could fsck or whatever.

Steve

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

The early superblocks that layout the filesystem geometry are probably gone. So you’ll need a tool that’s smart enough to search past those to later superblock backups that might still have the old HFS geometry. Then whatever data is left can be recovered.

If the inode table was completely destroyed you’re not going to have much luck. If you only lost some of the root inodes (that define the root directory etc), you may end up with a bunch of files in “lost+found” that need to be recovered.

I’d consider a call with the Diskwarrior guys just to discuss whether diskwarrior can do what I described

Steve

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

Those are kernel log files and they rotate regularly (kernel.log becomes kernel.log.1 etc).
They should not be so large. In fact, that’s an amazing size. They are text files! 420GB of text is amazing. That’s approximately 4 scientific libraries full of science books and journals (per file). with all of them together, your system is generating a library of congress or so each month 🙂

You probably can’t open them with a normal editor because of their size. You can probably go into a terminal and do this:

cd /var/log
sudo bash (to become root)
head -100 kernel.log

This will dump the first 100 lines of the file. I imagine something is spewing messages very rapidly. You need to figure out what that is and stop it.

Steve

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

Matt beat me to this.
Avid works just fine with a straight up NAS provided you aren’t capturing or rendering to the shared storage during the working day. That changes the index files which causes other users to reindex. Capture and render locally *or* setup shared scratch volumes and use those. Others can see them if necessary, but won’t typically have them mounted.

Since you want to use Pro Res, you’ll probably use Avid AMA anyhow (so you can stay in Pro Res and not have to ingest as MXF files). If this is the case, Avid works exactly like all the other apps. It treats it as a NAS volume.

For a small shop like you, I wouldn’t write off FCP X just yet either. It works well with Xsan and NAS volumes when you use the Sparse Disk Image method. It’s been written up a few times and it’s very elegant.

Steve Modica
Small Tree

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

Our experience has been that project files should not live on the shared storage.

This changes with FCP X and its new SAN capabilities. We’ve been testing Titanium as an iSCSI Xsan and FCP X will use it as a san volume. You can open project areas on the SAN volume and it correctly locks them so other FCP X users can’t come in and screw them up. It also lets you “detach” them so others can get to them.

This is not “volume” locking in the FibreJet sense. it’s directory locking. You can give any subdirectory on the SAN and it will lock it. It’s very nice.

Steve Modica
Small Tree

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

In my experience, memory and disk IO are more important that CPU speed. So what you might give up in CPU clock speed, make up for with more memory (perhaps from crucial because they are inexpensive) and some sort of external Thunderbolt Toaster raid that is meant for the purpose.

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

Your new Mac needs a really good “/” key cause you use it a *lot* 🙂

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

There is going to be a lot of hot plug and unplug weirdness for a while. When a PCIE bus just “goes away” the drivers are going to get errors. Most drivers will try to talk to the card to find out what’s wrong (oops! System hang). If they are smart and don’t try to talk to the card, then they have to tear down whatever was going on with the OS to free up the OS structures so the device can get plugged back in later. (otherwise, hotplug won’t work after an unplug). It’s going to take people a while.

Steve Modica
CTO, Small Tree Communications

[Steve Modica]

Ahem.. Small Tree’s is quite good.. and free. 🙂 Just check our downloads page.

Steve Modica
CTO, Small Tree Communications