fcp.co logo transparent
fcp clapperboard
Welcome, Guest
Username: Password: Remember me
25 Jan 2021
New boarders will have their posts moderated - Don't worry if you cannot see your post immediately.
Read More...
  • Page:
  • 1

TOPIC:

FCP share causing MAC os kernel crash ??? 12 Jul 2021 02:14 #115383

I have some projects that are "causing" Catalina 10.15.7 to crash when it gets close to completing a share to file. I am running FCP 10.5.2. I have been working on this project for a long time adding as I go along. The FCP bundle is 897 GB using one event and several projects using the library.

Today was the first time it happened and it is now consistently crashing the OS doing a share. I did add a lot of content to the project today. It crashes when it is about 85% complete.

My machine is an iMac 2017 w 32G and a 1TB HD and 400G free. I have two 6TB laCie drives one for data and the other a TM backup and there is plenty of space on each drive.

I tried to share another smaller project using another library and it worked fine. It seems to be related to this library because another project using this library crashed it as well. " Your computer was restarted because it encountered a problem". type message. Often I have to unplug the mac in order to restart.

I have run hdw diagnostics and recovery mode disk diagnostics on all drives and all is clean. I have not added any hdw or software. No plugins are installed and no new HDW has been installed.

Is there a potential correlation with this FCP library project and the crash? Is there a way to rebuild or trouble shoot it?

I deleted the preferences and the problem remains.

Thanks in advance
Rich

Please Log in to join the conversation.

Last edit: by ETdroneHome.

FCP share causing MAC os kernel crash ??? 12 Jul 2021 14:32 #115399

  • joema
  • joema's Avatar
  • Offline
  • Platinum Member
  • Platinum Member
  • Posts: 1918
  • Karma: 27
  • Thank you received: 444

ETdroneHome wrote: Catalina 10.15.7....running FCP 10.5.2...FCP bundle is 897 GB...Today was the first time it happened and it is now consistently crashing the OS doing a share. I did add a lot of content to the project today. It crashes when it is about 85% complete....iMac 2017 w 32G and a 1TB HD and 400G free. I have two 6TB laCie drives one for data and the other a TM backup and there is plenty of space on each drive....I tried to share another smaller project using another library and it worked fine. It seems to be related to this library because another project using this library crashed it as well. " Your computer was restarted because it encountered a problem". type message. Often I have to unplug the mac in order to restart...


This is a MacOS, hardware or system config issue. A user-layer app cannot crash the OS. It may seem to happen during a certain FCP task but that is incidental, and no fix to FCP source code will prevent it.

It is similar to a heavy truck crossing a defective bridge and then it collapses. The truck was within the weight limit of the bridge, but the structure was defective. E.g, the structural safety margin on the Brooklyn Bridge main cables is 600%. If it threatened to collapse, the solution would not be making lighter trucks but investigating and fixing the bridge.

Apple hardware diagnostics are very limited and an error-free pass means nothing. It is easy to run and if it finds something it is significant but no errors does not mean a "clean bill of health". The Apple Genius Bar has access to far more extensive overnight bench diagnostics which are more likely to find a hardware problem (assuming it exists).

A common cause of MacOS crashes is kernel-mode device drivers, which is why they have been deprecated for several years and app developers should have re-written those to use the newer, safer kernel extension API. You can check what kernel mode drivers you have using this terminal command:

kextstat | grep -v com.apple

I believe this command changed around Big Sur (can't remember about Catalina), and the new command is below; they both do the same thing:

kmutil inspect | grep -v com.apple

Please Log in to join the conversation.

FCP share causing MAC os kernel crash ??? 12 Jul 2021 15:21 #115402

Joema
Thanks for the response. I hear what you are saying and it makes sense. I asked Apple tech support (FCP) if Apple store techs have more extensive tests and he said they are the same. I questioned that. Before I lug this thing to an apple genius store I want to exhaust the other issues.

He suggested creating a new library and copying the project over to it and seeing if the results were the same. The results were identical. So the existing event seems to be ok since a new event was created.

So I was thinking of using this new copy of the project and creating several compound clips. Then I would create a project for each clip and export them to see if any of them pass. To use your analogy - where does the bridge fail? I am making a Super Duper bootable copy of the HD just to be safer.

I was also thinking of plugging the 6TB data TB drive into a 2013 mac running the same OS and FCP versions and see if that makes a difference. I made a fresh TM copy of it. I may first try to export it to the 2013 HD vs the LaCie Drive to rule out the external drive. I can open the project and made some small changes but did not save them and it seems to work. So the export process seems to be the issue but not sure how.

Yesterday I update the project adding clips, photos, text through out the project. However everything is in one event and this might isolate it.

I do not know what I would be looking at if I viewed kernel drivers, nor what do I compare it to. I have not upgraded to 10.5.4 or Big sur to avoid potential issues with this very important project.

This all just started maybe a bit on Saturday and certainly all Sunday. I worked on it all Sunday and it rendered all day.

Rich

Please Log in to join the conversation.

FCP share causing MAC os kernel crash ??? 15 Jul 2021 11:43 #115454

  • joema
  • joema's Avatar
  • Offline
  • Platinum Member
  • Platinum Member
  • Posts: 1918
  • Karma: 27
  • Thank you received: 444

ETdroneHome wrote: ... I asked Apple tech support (FCP) if Apple store techs have more extensive tests and he said they are the same. I questioned that. Before I lug this thing to an apple genius store I want to exhaust the other issues.


Their information is wrong. I have taken my machines to the Genius Bar several times and they ran overnight bench diagnostics on them. They don't like doing that so I had to encourage them. In one case it found a GPU problem on a 2015 iMac 27 that only showed up after an overnight run. There is a stark difference between being disinclined to run those diagnostics vs saying they don't exist.

ETdroneHome wrote: ....He suggested creating a new library and copying the project over to it and seeing if the results were the same. The results were identical. So the existing event seems to be ok since a new event was created...


He apparently does not understand the difference between kernel mode and user mode. In general you don't mess around with a user mode app trying to find a workaround to a MacOS crash. MacOS should never, ever crash due to a user mode app. If a regular user mode app could crash MacOS by accident, then a purpose-designed malicious app could do that easily, without using kernel-mode drivers or any special privileges.

You should be talking to a MacOS escalation support tech, not a Pro Apps person. IMO the moment the Pro Apps support person learned it was a MacOS crash, you should have been transferred to MacOS support.

The initial person may tell you they don't look at errorlogs, don't do any study of crash logs, etc. That is understandable, as it usually requires a "Tier 3" or higher support tech for that.

They will often suggest you backup, erase the machine and re-install all apps from scratch, not just re-load an image backup from Time Machine. While frustrating, that is understandable since it solves many problems and support engineers with the expertise to read crash logs are rare.

ETdroneHome wrote: ...I was also thinking of plugging the 6TB data TB drive into a 2013 mac running the same OS and FCP versions and see if that makes a difference. I made a fresh TM copy of it. I may first try to export it to the 2013 HD vs the LaCie Drive to rule out the external drive. I can open the project and made some small changes but did not save them and it seems to work. So the export process seems to be the issue but not sure how...


While there may be some benefit in verifying hardware configs and trying it on another machine, it is *extremely* unlikely that any software bug in FCP is responsible for a MacOS crash or that any software fix in FCP will prevent it.

In the case of a MacOS crash, the escalation support engineer needs to look at the crash log and do various searches in their support database for similar crash logs. For an Apple MacOS escalation support engineer, there is lots of info about this; some of it is even publicly documented:

"What caused that kernel panic?": eclecticlight.co/2020/06/17/what-caused-...he-log-in-diagnosis/

WWDC18: "Understanding Crashes and Crash Logs": developer.apple.com/videos/play/wwdc2018/414/ 

Analyzing a Crash Report: developer.apple.com/documentation/xcode/...yzing-a-crash-report 

Identifying the Cause of Common Crashes: developer.apple.com/documentation/xcode/...se-of-common-crashes 

Examining the Fields in a Crash Report: developer.apple.com/documentation/xcode/...ds-in-a-crash-report




ETdroneHome wrote: ...I do not know what I would be looking at if I viewed kernel drivers, nor what do I compare it to. I have not upgraded to 10.5.4 or Big sur to avoid potential issues with this very important project...


You don't need to understand the specifics of the terminal command listing 3rd-party kernel mode drivers. You just make note of the output. You could copy/paste the result in a post here -- it should be fairly brief. You could then run the same command on the 2013 Mac and look for any differences -- assuming that machine does not crash.

For a MacOS crash, the area to focus on is hardware, system config, kernel mode drivers, etc -- not the application layer.

Please Log in to join the conversation.

Last edit: by joema.

FCP share causing MAC os kernel crash ??? 22 Jul 2021 19:04 #115513

After a lot of effort I decided to take the iMac to a Genius Bar and they determined it was the memory that was failing. They offered to replace the 32G with official Apple memory at a cost of $600.

Fortunately it was OWC memory and they replaced it under warranty.

Thanks to joema for the help.

Rich

Please Log in to join the conversation.

FCP share causing MAC os kernel crash ??? 24 Jul 2021 11:18 #115526

  • joema
  • joema's Avatar
  • Offline
  • Platinum Member
  • Platinum Member
  • Posts: 1918
  • Karma: 27
  • Thank you received: 444
Thanks for posting that conclusion. A few years ago I had a recurring FCP crash with similar stack trace on my 2017 iMac 27. FCP version was 10.4.1 through 10.4.6. It would more frequently happen when I was doing color correction with scopes up. I never figured it out and had just gotten a 2017 iMac Pro, so I switched to that and it never happened on that machine -- despite running the same version of MacOS, FCP and same FCP libraries with same media.

My 2017 iMac 27 had 32GB of Apple RAM, but I suspect it may have been an intermittent RAM problem (like yours which was found by the Genius Bar on OWC RAM). It's likely not that OWC RAM was poor, but any RAM can fail, sometimes intermittently.

The Mac Pro and iMac Pro have ECC RAM which is transparently protected from single-bit failures. Multi-bit failures will cause a problem but those will be logged, not slip by as a mysterious crash. I ran Apple Diagnostics on the 2017 iMac 27, but (as usual) it never found anything and I didn't take it to the Genius Bar for overnight bench diagnostics.

For Mac Pros and iMac Pros with ECC RAM, you can inspect if any ECC errors were logged by looking at System Report>Hardware>Memory, and inspecting the status column. It will say OK if no errors, or "ECC Errors" if any have occurred on that module. If any ECC errors are seen on an iMac Pro or Mac Pro, rigorous RAM diagnostics should be run and maybe the module replaced.

While there are GPU functions in both crash logs, I don't think it's a GPU error but rather certain operations such as color correction may cause rapid data transfer between RAM and GPU VRAM, which could stress any weakness in system RAM, eventually resulting in reading wrong data which produces a crash.

For future reference, below is one of your stack traces from a FCP 10.5.3 crash, followed by one of mine on 10.4.1. (unable to format them properly for in-line insertion; using attachment).
Attachments:

Please Log in to join the conversation.

Last edit: by joema.
  • Page:
  • 1