Media & Cache on shared storage network

Wondering where you all keep your Media and Cache folders when working in a shared network environment. We have an EVO with multiple users. All has been working amazingly until recently. We keep getting a FCP dialog that says that FCP cannot save changes to the Library. The storage location for one or more libraries is no longer available. FCP then closes that library but doesn’t crash. In fact all remaining libraries are still active. Our setup has been media , cache and all video, graphics, sound etc kept on a share of the network. Libraries and BU’s on a different share of the network. This worked well for over a year then all of a sudden this is popping up.

Anything change? What software version? What OS?

It started when we were on High Sierra and the highest version of FCP. We since we have upgraded to Mojave and FCP 10.4.8. No difference. One little test we did was move the library to a direct attached Pegasus drive and that ran smoothly. When I moved the library back to an SMB network share the problem came back.
I was thinking that perhaps the storage location of backups, cache and or media might be a factor.

You should not have any issues storing FCP X Libraries, cache folders and media on an SMB share or on separate shares. We do this all the time on a Jellyfish. This works with any MacOS or FCP X version. So I suggest you contact EVO support, they need to fix this.

- Ronny

Thanks Ronny,
Agreed it should work and has been working well for some time. I need to figure what has changed or causing conflict. The other thing is I moved the library off the EVO onto another SMB network and had the same problem.
It’s really a strange issue as none of the shares dismount or loose connection. This is a tough one.

It is worth noting that there have been some major changes to SMB in the latest 2 MacOS updates. But that should not cause the issue you describe. I have a major client in Norway who works with more than 70 fcp x workstations over smb and they don’t have any issues at all. Have you tried storing the libraries on the same share as the media and cache folders?

- Ronny

I guess that’s what’s so frustrating g about this issue. I know it works and works well also. Just some thing went wrong somewhere.
I have not tried keeping everytjing on the same share. I thought it would be best to keep them separate. I’ll give that a try tomorrow.

Moved the Library to the same share as Media & Cache and got the same fault.

I have talked with our tech team about this. As I said, there have been big changes in the way MacOS handles SMB since High Sierra. We were aware of these changes and pushed updates to all our clients prior to the OS updates to prevent any such issues. So you really should contact EVO to see if there are any updates you should do.

- Ronny

Thanks again Ronny. I have been working with SNS since the beginning on this and we can't exactly recreate the problem consistently at all. As far as SMB goes they are aware of packet signing issues with performance but no direct connection to this issue to pinpoint a patch of any kind.

I’m having some trouble as well. When I move libraries off of the share, that library somehow gets “corrupted”, but this is exclusive to Catalina. When playing back in the timeline, the share is ejected and fcpx gets the warning about not being able to save. It’s not just the share being ejected, but the whole 10Gb pipe is shut down and then subsequently recovers. Internet and everything goes away and then comes back. That library is then inoperable on Catalina based machines, but opens and plays just fine on Mojave machines.

If start a new library using Catalina, and I don’t move the library off of the share, then it works. Seems to be some sort of ‘bug’ in Catalina. I have more testing to do to really track it down, but that’s what I’ve learned so far.

I had our workflow team do some testing using exactly the same configurations and processes. I can certify that we cannot reproduce the issues on a Jellyfish NAS.

- Ronny

Thank you Ronny for digging in and looking into this. As I’ve mentioned the intermittent nature of the issue is problematic to troubleshoot. I can have a library open all day without one issue. Then open it another day and three to four issues in the AM then be fine the rest of the day. I’m about to setup another session with SNS to help look into what’s going on.
Not sure it’s entirely on them. I’m trying to test having the library stored on another SMB share separate from EVO.

What kind of Ethernet adapter do you have?

Solarflare 10Gbe in a Sonnet expansion chassis.

Dealing with a very similar issue with our system. Here's where we're at as we troubleshoot with Apple and Lumaforge.

Issue:
Loading a timeline in FCPX 10.4.8, on Mac OS 10.15.3, forces the server connection to our Lumaforge Jellyfish Rack SMB and NFS shares to suddenly disconnect. FCPX then prompts with a "Storage disconnected" error, forcing the library to close. SMB and NFS shares then need to be remounted in the JellyConnect app.

Systems Affected:
Any computer running 10.15.3.

Systems unaffected:
Any computer running 10.14.6 or earlier.

We have had no issues with the exact same assets or libraries on any system running 10.14.6 or earlier, on FCPX 10.4.8 or earlier, connected to internal or external assets. We also have no issues with the exact same assets and libraries running off the internal drive of a 10.15.3 system. Issue is only present when using external SMB or NFS based assets and libraries.

We have troubleshooted ethernet adaptors, CAT 6/7 cables, different server ports, different assets, different 2 year old libraries, updated JellyFish etc etc.

Currently this is sitting with Lumaforge and Apple, but effectively we are unable to use any newly released Apple computer that can't be downgraded to 10.14.6. Which means no new Mac Pro or MacBook Pro 16".

If anyone has any ideas, we're running out of ideas here!

You can’t run 10.4.6 in the current OS?

All,

There is a confirmed issue between Catalina and FCP X when using SMB or NFS protocols. The weird thing is that it does not happen on every system that is running this combination, which makes it very difficult to troubleshoot. As I said, we have quite a few clients running Catalina and FCP X on a Jellyfish without any issues. But recently we also have received reports as the one from Stefan.

This is an incompatibility issue between Catalina and FCP X over a network, running very specific hardware combinations. We have reported this directly to Apple (as Stefan has seen on our support portal) and are waiting for a fix. We have been trying to work around the issue. But as far as I see now, the fix needs to come from Apple.

In the meantime, we advise NOT to use Catalina with FCP X on a network storage. In fact, you never should upgrade an OS without first checking if there are any reported issues with the applications and the hardware you use. The big bummer, however, is that all new MBPs andMPs come with Catalina installed and it's not really easy to downgrade to Mojave. But there are hundreds of pages on the internet that explain how to downgrade from Catalina, so a lot of people must be doing it.

- Ronny

Thanks for all the info. This is great to know as we place orders for MP’s. On Catalina. I was under the impression you couldn’t dial them back to Mojave.
Oddly my behavior started in High Sierra and into Mojave. Perhaps there still is a patch needed. I have several days in a particular library where everything works as it should then can have 3-4 can’t save messages within 15minutes. Then totally fine again.
In working with SNS they said we've found a way to consistently reproduce stalled communication with SMB targets when working with FCP X. I’ll check in with them and see where we are at.

Thanks for the info everyone.

As discouraging as it is, I’m glad to know that it can be reproduced elsewhere. I hope a fix can come swiftly as it’s really troublesome not being able to use new hardware.

oldsoul, we have similar setups and experiences but Mojave is by far more stable as the connection can be recovered in an instant, while on Catalina, the connection recovers very slowly. New Catalina based machines are unusable at this point and it’s extremely frustrating.