Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

bax2023

macrumors regular
Nov 14, 2023
124
158
Serbia
Now 7 days. Yes, it hasn't been long, but so far everything looks good.

Its quite enough to start testing. I have Sonoma 14.5 on one of the SSD drives so I'll give it a try.

EDIT: After more than 2 months without a signle issue with Sequoia betas, I booted Sonoma 14.5 to update it to 14.6.1 and test new version, and both of my NVMes vanished.

No matter what anybody writes here, this is up to OS 100%. Nothing to do with hardware or compatibility or anything else.

OK, lets update to 14.6.1 and maybe confrm findings by @Rimmsi.

Screenshot 2024-08-12 at 19.01.54.png
 
Last edited:
  • Like
Reactions: danano

bax2023

macrumors regular
Nov 14, 2023
124
158
Serbia
macOS Sequoia 15.0 Beta 6 (24A5320a) upgraded the firmware versions:

FROM:
System Firmware Version
: 2069.0.0.0.0 (iBridge: 22.16.10339.5.2,0)
OS Loader Version: 582~883

TO:
System Firmware Version
: 2069.0.0.0.0 (iBridge: 22.16.10350.5.1,0)
OS Loader Version: 582~1070


And still no sign of the mount issues 🥳
 

phrehdd

Contributor
Oct 25, 2008
4,496
1,455
I found a work around! I think there's something going on with Apple RAID and OS Monterey and higher. I read some thread about APFS having issues with external drives, so I tried reformatting the drives in HFS+, but that didn't work. The system would see the newly formatted RAID fine, but as soon as I restarted the machine, the RAID would disappear and couldn't be remounted.

So the fix is using SoftRaid. It changes the Apple Raid to a SoftRaid array. It doesn't destroy any data. There are two drawbacks: 1) You have to always run SoftRaid in the background and 2) you have to purchase it. But it fixed both systems and there are no more issues with the RAIDs missing in both systems.

I would hope Apple will fix this in future releases, but since it's still happening in Ventura, I doubt it.
Bigger issue if I recall, Softraid now is a subscription model.
 

bcomer

macrumors regular
Jan 25, 2008
205
145
Ottawa
macOS Sequoia 15.0 Beta 6 (24A5320a) upgraded the firmware versions:

FROM:
System Firmware Version
: 2069.0.0.0.0 (iBridge: 22.16.10339.5.2,0)
OS Loader Version: 582~883

TO:
System Firmware Version
: 2069.0.0.0.0 (iBridge: 22.16.10350.5.1,0)
OS Loader Version: 582~1070


And still no sign of the mount issues 🥳
I am on 14.6.1 and needed to reboot twice to get my SSDs back.
 

CanPhantom

macrumors member
Oct 21, 2014
48
26
I am on 14.6.1 and needed to reboot twice to get my SSDs back.

Yep ditto I spoke way too soon. They hooped it again in 14.6.1, right back to as bad as it was with 14.5. Ah well, it was a nice few weeks without this bug I guess...
 

bax2023

macrumors regular
Nov 14, 2023
124
158
Serbia
I can also confirm that 14.6.1 didn't address this issue. After 4th reboot, NVMe haven't mounted.
BTW all is still well for me in Sequoia.
 

doobydoooby

macrumors regular
Oct 17, 2011
245
355
Genève, Switzerland
I've had a lot of success with a piece of software called Jettison (https://www.stclairsoft.com/Jettison/). It ejects everything before the computer is turned off. It's not 100% perfect, but I would say it has reduced the number of times that the Mac no longer boots to a handful; I reckon it's reduced the failure rate by about 90%. It also allows me to effectively use sleep, which I never could before because of the likelihood of failure upon waking.

I know a few of you have had mixed results with manually ejecting disks before shutdown, so perhaps this isn't the only issue going on, but I thought it worth highlighting since it's been a great success for me. I have three Samsung Evo SSDs mounted on a Sonnet sled and am running Sonoma 14.5. Before using Jettison, it typically took two or three attempts to get the computer to boot successfully without either a kernel panic or a failure to mount those external drives. These days, failure only happens once per fortnight or so, and that's been my consistent experience over the last two or three months, so it's stable. I turn 'pooter off daily.
 

H2SO4

macrumors 603
Nov 4, 2008
5,828
7,103
Zero cards thar require Bifurcation work with Mac Pro because the machine itself does not support it. Only the first drive will be recognized from those cards. Thats why Mac Pro requires cards with PEX PCIe switch so that it will do the "Bifurcation" role and comminicate between PCIe bus and all NVMe drives separately.

My guess is that most of the problems are due to misscommunication between MP PCIe bus and the card with those chips.

Sure, there are RAID cards which don't use PCIe switch, but those are MUCH more expensive than OWC or Sonnet, but must users in this thread have OWC/Sonnet combination or HighPont with PCIe switch.
Question I have to that is why does that same problem never manifest when running Windows ?
 

H2SO4

macrumors 603
Nov 4, 2008
5,828
7,103
My guess is that Windows is not having entire subsystems replaced with "equivalents" from a cellphone & tablet operating system.

macOS is going to end up as a UI theme, I suspect.
So is it safe to say that it's the Mac OS that doesn't support bifurcation and not the MacPro?.
macOS is going to end up as a UI theme, I suspect.
Ok, that's quite funny.
 

tsialex

Contributor
Jun 13, 2016
13,454
13,601
Intel Lane Partitioning, the correct name for bifurcation, is a hardware + firmware resource - you need both to make it work.

MacPro7,1 have a chipset that supports the hardware part, but not the firmware, so, bifurcation is not supported even with Windows/Linux.

Even if Apple developed the firmware part to enable Intel Lane Partitioning, it would only work for the PCIe slots 1 & 3 that are directly connected to the CPU, the ones that are used for the MPX GPUs, so, a no-go from the start.

When you install a multiple NMVe card that works with Intel Lane Partitioning to the MacPro (5,1 or 7,1, same behavior happens with both) you only access the first blade, completely ignoring the other blades.
 
  • Like
Reactions: H2SO4 and mattspace

H2SO4

macrumors 603
Nov 4, 2008
5,828
7,103
Intel Lane Partitioning, the correct name for bifurcation, is a hardware + firmware resource - you need both to make it work.

MacPro7,1 have a chipset that supports the hardware part, but not the firmware, so, bifurcation is not supported even with Windows/Linux.

Even if Apple developed the firmware part to enable Intel Lane Partitioning, it would only work for the PCIe slots 1 & 3 that are directly connected to the CPU, the ones that are used for the MPX GPUs, so, a no-go from the start.

When you install a multiple NMVe card that works with Intel Lane Partitioning to the MacPro (5,1 or 7,1, same behavior happens with both) you only access the first blade, completely ignoring the other blades.
Ok gotcha, that pretty much fills the hole I had in my understanding.
Thankyou.
 

H2SO4

macrumors 603
Nov 4, 2008
5,828
7,103
Just to be clear, you guys that are running Sequoia without problems are you all actually booting from it or is it just a storage drive?
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.