MP All Models PCIe SSDs - NVMe & AHCI

[fatespawn]

Any suggestions? I can't format my 970EVO Plus. It's a 500GB revision 1B2QEXM7

I've gotten a few different errors. (Error 83?) and I went through the "unable to unmount" difficulty. Finally, was able to unmount with sudo diskutil unmount force.. So, that got it unmounted after the first failed initialization, but now I get "unable to write to the last block of the device" both in the command line and in the GUI.

Suggestion?

Oh yeah, I'm using the Aquacomputer Kryo M.2 Adapter.

 

[AidenShaw]

Any suggestions? I can't format my 970EVO Plus. It's a 500GB revision 1B2QEXM7

I've gotten a few different errors. (Error 83?) and I went through the "unable to unmount" difficulty. Finally, was able to unmount with sudo diskutil unmount force.. So, that got it unmounted after the first failed initialization, but now I get "unable to write to the last block of the device" both in the command line and in the GUI.

Suggestion?

Oh yeah, I'm using the Aquacomputer Kryo M.2 Adapter.

I hope that command/utility isn't writing zeroes to the whole drive - that's a bad thing to do on an SSD. It adds a full wear cycle, and makes the visible disk space 100% full.

You want an SSD utility that does a "secure erase". Secure erase electronically deletes all of the data and make the disk 0% full, without any wear and tear.

 

[fatespawn]

I hope that command/utility isn't writing zeroes to the whole drive - that's a bad thing to do on an SSD. It adds a full wear cycle, and makes the visible disk space 100% full.

You want an SSD utility that does a "secure erase". Secure erase electronically deletes all of the data and make the disk 0% full, without any wear and tear.

Thanks,

Didn't matter anyway. My web searches yielded "cabling issues" so, I pulled the card, reinserted the stick, re-racked the card, and all seems well after the reboot.

*edit* well I take that back. Still trying to figure out what is going wrong. The format went well, but now MacOS is "reporting errors" Maybe a bad drive. I'll have to keep playing.

 

[AidenShaw]

Didn't matter anyway.

It does matter.

If you wrote to the entire disk, it is 100% full.

When you eventually find the problem, it will still be 100% full until you do a secure erase or force TRIM on the disk.

 

[sailmac]

Still trying to figure out what is going wrong. The format went well, but now MacOS is "reporting errors" Maybe a bad drive.

Have you looked at S.M.A.R.T. details for the drive? From experience DriveDx is comprehensive and may reveal more specifically what's going on with your drive.

 

[fatespawn]

You misunderstand. It doesn't matter because I can't even format the thing. It's been doing this for 5 minutes. Probably a little long, right?

[doublepost=1551578049][/doublepost]

Have you looked at S.M.A.R.T. details for the drive? From experience DriveDx is comprehensive and may reveal more specifically what's going on with your drive.

Thanks - yes. I tried DriveDx. It doesn't even show up in the list of drives.

 

[sailmac]

Thanks - yes. I tried DriveDx. It doesn't even show up in the list of drives.

Does it show up in Terminal if you

diskutil list

 

[HunPro]

I put the Mac Pro to sleep for the first time, and after a while it woke up, crashed, restarted and showed a problem report on startup:

Sun Mar  3 13:39:42 2019

*** Panic Report ***
panic(cpu 8 caller 0xffffff7f8e089f4a):
nvme: "InitializeNVMe error =
 0xE00002D7\n"@/BuildRoot/Library/Caches/com.apple.xbs/Sources/IONVMeFamily/IONVMeFamily-356.71.1/IONVMeController.cpp:304

IONVMeController.cpp

High Sierra, 140.0.0.0.0

I put it to sleep around 11:39. Somehow it woke up 2 hours later and crashed. It seems to be NVMe related.

I would like to put it to a lower power mode when unused, but I don't need anything fancy. I'm not sure why it woke up. The Mac Pro should not have Power Nap, right?

Anyone experienced such a thing?

One solution that might be applicable from a Macbook NVMe topic:

It's only on battery that the hibernation normaly occurs after 3 hours and the mac can't wake up and read the Ram data from the hibernation file.
And this can be prevented by desactivating the deep sleep mode, with typing "sudo pmset -a standby 0" in the terminal.

Also

sudo pmset -a hibernatemode 0

I don't need hibernation really, the RAM can stay powered, so this should be fine.

 

[tsialex]

I put the Mac Pro to sleep for the first time, and after a while it woke up, crashed, restarted and showed a problem report on startup:

Sun Mar  3 13:39:42 2019

*** Panic Report ***
panic(cpu 8 caller 0xffffff7f8e089f4a):
nvme: "InitializeNVMe error =
 0xE00002D7\n"@/BuildRoot/Library/Caches/com.apple.xbs/Sources/IONVMeFamily/IONVMeFamily-356.71.1/IONVMeController.cpp:304

IONVMeController.cpp

High Sierra, 140.0.0.0.0

I put it to sleep around 11:39. Somehow it woke up 2 hours later and crashed. It seems to be NVMe related.

I would like to put it to a lower power mode when unused, but I don't need anything fancy. I'm not sure why it woke up. The Mac Pro should not have Power Nap, right?

Anyone experienced such a thing?

What's your NVMe blade/adapter? It's a clean install?

This is a KP a lot more common with Retina MacBooks Pro that had the Apple OEM SSD changed to a NVMe blade, did you cloned your rMBP to your Mac Pro?

 

[HunPro]

What's your NVMe blade/adapter? It's a clean install?

This is a KP a lot more common with Retina MacBooks Pro that had the Apple OEM SSD changed to a NVMe blade, did you cloned your rMBP to your Mac Pro?

I can't tell you without disassembly. It has a blue and a red led. The SSD is a Toshiba xg4 oem blade.

It's a clean install.

 

[tsialex]

I can't tell you without disassembly. It has a blue and a red led. The SSD is a Toshiba xg4 oem blade.

It's a clean install.

XG4 is one of the first OEM NVMe drives, contemporaneous of Samsung 950 Pro that we know to be Mac Pro incompatible. Blades of this era have lots of quirks when installed into a Mac Pro.

If you continue to have problems, maybe it will be wise to exchange it.

 

[HunPro]

XG4 is one of the first OEM NVMe drives, contemporaneous of Samsung 950 Pro that we know to be Mac Pro incompatible. Blades of this era have lots of quirks when installed into a Mac Pro.

If you continue to have problems, maybe it will be wise to exchange it.

I will, but first I try the disabled hibernation sleep. I can put up with some quirks

UPDATE: Did a quick sleep / wake, came back just fine. Hoping that hibernation is the problem, I can live without it.

 

[tsialex]

I will, but first I try the disabled hibernation sleep. I can put up with some quirks

UPDATE: Did a quick sleep / wake, came back just fine. Hoping that hibernation is the problem, I can live without it.

Just to you know, the XG4 brother (same NAND and controller, different firmware) that Plextor sold at the time are not compatible with Mac Pro, not even show itself into System Information.

 

[HunPro]

Just to you know, the XG4 brother (same NAND and controller, different firmware) that Plextor sold at the time are not compatible with Mac Pro, not even show itself into System Information.

This one does, and I could install Mac OS on it just fine.

Device Name:    THNSF5256GPUK TOSHIBA
Media Name:    AppleAPFSMedia
Medium Type:    SSD
Protocol:    PCI-Express
Internal:    No

Sure, it says external.

Having the RAM powered during sleep is not a deal breaker, this is not a portable. Hell, it means faster wake up and less wear on the SSD when it tries to write up to 64 gigs during hibernation.

Every day I can live with it is a day that the SSDs on the market get cheaper and faster, so I chose frugality over comfort for now, if this stays stable, and upgrade when there's a good deal.

It's a used Mac so the owner before used this for a while. 10 of these were for sale with a similar config, so I guess it should work with a few restrictions on the long run, if it did for the previous owners.

 

[tsialex]

This one does, and I could install Mac OS on it just fine.

Device Name:    THNSF5256GPUK TOSHIBA
Media Name:    AppleAPFSMedia
Medium Type:    SSD
Protocol:    PCI-Express
Internal:    No

Sure, it says external.

Having the RAM powered during sleep is not a deal breaker, this is not a portable. Hell, it means faster wake up and less wear on the SSD when it tries to write up to 64 gigs during hibernation.

Every day I can live with it is a day that the SSDs on the market get cheaper and faster, so I chose frugality over comfort for now, if this stays stable, and upgrade when there's a good deal.

It's a used Mac so the owner before used this for a while. 10 of these were for sale with a similar config, so I guess it should work with a few restrictions on the long run, if it did for the previous owners.

All PCIe drives are external to the Mac Pro firmware.

Not related to the KP, but since it's an used blade and it's a SED model, maybe you could do a secure erase with Windows, just to be sure that all cells are fresh.

Check the state of it with DriveDX, btw.

 

[fatespawn]

Does it show up in Terminal if you

diskutil list

Yes. It showed up as Disk0. I tried to erase and format via the command line with no luck. Eventually, I was able (after a reboot or two and reseating the PCIe card) format it. I thought all was well. But when I attempted to write to it, Carbon Copy Cloner reported a ton of write errors and only 20 of 200GB copied.

I might try a different PCI slot today to see if that might be the problem.

 

[HunPro]

All PCIe drives are external to the Mac Pro firmware.

Not related to the KP, but since it's an used blade and it's a SED model, maybe you could do a secure erase with Windows, just to be sure that all cells are fresh.

Check the state of it with DriveDX, btw.

Excellent tip, just did that. Some parts of the report seem to be bogus:

Advanced SMART Status : OK
Overall Health Rating : GOOD 100%
SSD Lifetime Left Indicator : GOOD 100%
Issues found : 0

Volumes : MacProSSD
Device Path : /dev/disk0
Total Capacity : 256.1 GB (256 060 514 304 Bytes)
Model Family : Toshiba NVMe SSD
Model : THNSF5256GPUK TOSHIBA
Firmware Version : 51025KLA
Drive Type : SSD

Power On Time : 61 hours (2 days 13 hours) (not a chance!)
Power Cycles Count : 73 (plausible, the HDD has around 170 cycles)
Current Power Cycle Time : 2.4 hours

=== PROBLEMS SUMMARY ===
Failed Indicators (life-span / pre-fail) : 0 (0 / 0)
Failing Indicators (life-span / pre-fail) : 0 (0 / 0)
Warnings (life-span / pre-fail) : 0 (0 / 0)
I/O Error Count : 0 (0 / 0)


=== IMPORTANT HEALTH INDICATORS ===
ID NAME RAW VALUE STATUS
7 Data Units Written 1 432 575 (683.1 GB) 100% OK
14 Media and Data Integrity Errors 0 100% OK

=== DRIVE ERROR LOG ===
error log is empty

 

[tsialex]

Excellent tip, just did that. Some parts of the report seem to be bogus:

PowerOnTime seems really weird, the rest seems ok for an almost new blade.

 

[HunPro]

PowerOnTime seems really weird, the rest seems ok for an almost new blade.

I suspect that these Mac Pros were used in an office, all 10+ of them, with a similar config and a specific use case. With 16 GBs of memory, little swapping, 700GBs of total writes is plausible.

However, I ran this on my MacBook with a 10 day uptime. The writes are at 1 TB at the moment, and 42.1 TBs (!) during its 29 month lifetime. Sure, I'm a heavy user and always have a multiple gigabyte swap file. I suspect that the read/write amount being so close (46.5 /42.1 TB) points to this traffic being mostly swapping + hibernation (it's a laptop after all with 16 GBs of ram).

Wear level is reported to be 80%.

Advanced SMART Status : OK
Overall Health Rating : GOOD 100%
SSD Lifetime Left Indicator : GOOD 100%
Issues found : 0

Volumes : SSD 256 GB
Device Path : /dev/disk0
Total Capacity : 251.0 GB (251 000 193 024 Bytes)
Model Family : Apple (Samsung-based) SSDs
Model : APPLE SSD SM0256F
Firmware Version : UXM2JA1Q
Drive Type : SSD

Power On Time : 20 961 hours (29 months 3 days 9 hours)
Power Cycles Count : 21 184
Current Power Cycle Time : 240.6 hours

=== CURRENT POWER CYCLE STATISTICS ===
Data Read : 1.0 TB
Data Written : 870.8 GB

=== PROBLEMS SUMMARY ===
Failed Indicators (life-span / pre-fail) : 0 (0 / 0)
Failing Indicators (life-span / pre-fail) : 0 (0 / 0)
Warnings (life-span / pre-fail) : 0 (0 / 0)
Recently failed Self-tests (Short / Full) : 0 (0 / 0)
I/O Error Count : 0 (0 / 0)

=== IMPORTANT HEALTH INDICATORS ===
ID NAME RAW VALUE STATUS
5 Retired Block Count 0 100% OK
173 Wear Leveling Count 0x131026101C5 80.0% OK
175 Host Writes MiB 44 124 199 (42.1 TB) 99.0% OK
192 Unsafe Shutdown Count 89 99.0% OK
197 Current Pending Block Count 0 100% OK
199 UDMA CRC Error Count 0 100% OK


5 Retired Block Count Pre-fail online 0
197 Current Pending Block Count Life-span online 0 100 0 100 2019. 03.
199 UDMA CRC Error Count Life-span online 0 200 0 200 2019. 03.

=== DRIVE ERROR LOG ===
error log is empty

=== DEVICE STATISTICS ===

[0x01] General Statistics (revision 2)
0x08 Lifetime Power-on Resets : 21 184
0x10 Power-on Hours : 20 961 hours
0x18 Logical Sectors Written : 90 366 364 183 (42.1 TB)
0x20 Number of Write Commands : 970 370 881
0x28 Logical Sectors Read : 99 928 610 509 (46.5 TB)

[0x04] General Errors Statistics (revision 1)
0x08 Number of Reported Uncorrectable Errors : 0
0x10 Resets Between Cmd Accept. and Compl. : 0

[0x07] Solid State Drive Statistics (revision 1)
0x08 Percentage Used Endurance Indicator : 0 %

 

[fatespawn]

Mac-Pro:~ xxxxxxx$ diskutil partitiondisk disk2 1 gpt apfs "nvme" 0
Started partitioning on disk2
Unmounting disk
Creating the partition map
Error: -69760: Unable to write to the last block of the device
Mac-Pro:~ xxxxxx$

Well, tried a different PCI slot. No changes. I guess I'll send the drive back. Could be the kryoM.2 as well, but I'll wait until 2 blades don't work before I go there.

 

[tsialex]

Mac-Pro:~ xxxxxxx$ diskutil partitiondisk disk2 1 gpt apfs "nvme" 0
Started partitioning on disk2
Unmounting disk
Creating the partition map
Error: -69760: Unable to write to the last block of the device
Mac-Pro:~ xxxxxx$

Well, tried a different PCI slot. No changes. I guess I'll send the drive back. Could be the kryoM.2 as well, but I'll wait until 2 blades don't work before I go there.

Did you try removing the partition block with sudo gpt destroy?

diskutil list
diskutil unmountdisk diskXX
sudo gpt destroy diskXX

 

[fatespawn]

Did you try removing the partition block with sudo gpt destroy?

diskutil list
diskutil unmountdisk diskXX
sudo gpt destroy diskXX

I think I did last night. It was late. I'll try again. I rebooted - it seemed a normal restart - but I got the "your computer was restarted because of a problem" message. This is from the log:

NVMeController.....?

*** Panic Report ***
panic(cpu 10 caller 0xffffff7f98f9d277): nvme: "systemWillShutdown, ShutdownNVMe() completed with status = 0xe00002e9\n"@/BuildRoot/Library/Caches/com.apple.xbs/Sources/IONVMeFamily/IONVMeFamily-387.230.1/IONVMeController.cpp:532
Backtrace (CPU 10), Frame : Return Address
0xffffff82cf653c40 : 0xffffff80171aeb0d
0xffffff82cf653c90 : 0xffffff80172e8653
0xffffff82cf653cd0 : 0xffffff80172da07a
0xffffff82cf653d40 : 0xffffff801715bca0
0xffffff82cf653d60 : 0xffffff80171ae527
0xffffff82cf653e80 : 0xffffff80171ae373
0xffffff82cf653ef0 : 0xffffff7f98f9d277
0xffffff82cf653f30 : 0xffffff80178bcf4f
0xffffff82cf653f70 : 0xffffff80178bccc8
0xffffff82cf653fa0 : 0xffffff801715b0ce
Kernel Extensions in backtrace:
com.apple.iokit.IONVMeFamily(2.1)[ABA6DFDD-AC72-3008-83FE-EF9722E66FEB]@0xffffff7f98f8f000->0xffffff7f98fcefff
dependency: com.apple.driver.AppleMobileFileIntegrity(1.0.5)[A2492141-9480-3177-BC7D-455925BB4A2E]@0xffffff7f97f17000
dependency: com.apple.iokit.IOPCIFamily(2.9)[5C9A453F-559B-3683-9E81-D288D13A33CE]@0xffffff7f97a95000
dependency: com.apple.driver.AppleEFINVRAM(2.1)[328FCBD7-8C2C-3A4A-AF92-9270C1403891]@0xffffff7f980ed000
dependency: com.apple.iokit.IOStorageFamily(2.1)[9B2E7108-AA17-3998-BF41-4B1297F455E5]@0xffffff7f97d7d000
dependency: com.apple.iokit.IOReportFamily(47)[5F165AE5-F4F2-3415-857C-34F2462A730E]@0xffffff7f98489000
[doublepost=1551635135][/doublepost]

Mac-Pro:~ xxxxxxx$ diskutil unmount disk2
disk2 was already unmounted or it has a partitioning scheme so use "diskutil unmountDisk" instead
Mac-Pro:~ xxxxxxx$ diskutil unmountDisk disk2
Unmount of all volumes on disk2 was successful
Mac-Pro:~ xxxxxxx$ sudo gpt destroy disk2
Password:
gpt destroy: unable to open device 'disk2': Device error

Failed.

 

[tsialex]

I think I did last night. It was late. I'll try again. I rebooted - it seemed a normal restart - but I got the "your computer was restarted because of a problem" message. This is from the log:

NVMeController.....?

*** Panic Report ***
panic(cpu 10 caller 0xffffff7f98f9d277): nvme: "systemWillShutdown, ShutdownNVMe() completed with status = 0xe00002e9\n"@/BuildRoot/Library/Caches/com.apple.xbs/Sources/IONVMeFamily/IONVMeFamily-387.230.1/IONVMeController.cpp:532
Backtrace (CPU 10), Frame : Return Address
0xffffff82cf653c40 : 0xffffff80171aeb0d
0xffffff82cf653c90 : 0xffffff80172e8653
0xffffff82cf653cd0 : 0xffffff80172da07a
0xffffff82cf653d40 : 0xffffff801715bca0
0xffffff82cf653d60 : 0xffffff80171ae527
0xffffff82cf653e80 : 0xffffff80171ae373
0xffffff82cf653ef0 : 0xffffff7f98f9d277
0xffffff82cf653f30 : 0xffffff80178bcf4f
0xffffff82cf653f70 : 0xffffff80178bccc8
0xffffff82cf653fa0 : 0xffffff801715b0ce
Kernel Extensions in backtrace:
com.apple.iokit.IONVMeFamily(2.1)[ABA6DFDD-AC72-3008-83FE-EF9722E66FEB]@0xffffff7f98f8f000->0xffffff7f98fcefff
dependency: com.apple.driver.AppleMobileFileIntegrity(1.0.5)[A2492141-9480-3177-BC7D-455925BB4A2E]@0xffffff7f97f17000
dependency: com.apple.iokit.IOPCIFamily(2.9)[5C9A453F-559B-3683-9E81-D288D13A33CE]@0xffffff7f97a95000
dependency: com.apple.driver.AppleEFINVRAM(2.1)[328FCBD7-8C2C-3A4A-AF92-9270C1403891]@0xffffff7f980ed000
dependency: com.apple.iokit.IOStorageFamily(2.1)[9B2E7108-AA17-3998-BF41-4B1297F455E5]@0xffffff7f97d7d000
dependency: com.apple.iokit.IOReportFamily(47)[5F165AE5-F4F2-3415-857C-34F2462A730E]@0xffffff7f98489000
[doublepost=1551635135][/doublepost]

Mac-Pro:~ xxxxxxx$ diskutil unmount disk2
disk2 was already unmounted or it has a partitioning scheme so use "diskutil unmountDisk" instead
Mac-Pro:~ xxxxxxx$ diskutil unmountDisk disk2
Unmount of all volumes on disk2 was successful
Mac-Pro:~ xxxxxxx$ sudo gpt destroy disk2
Password:
gpt destroy: unable to open device 'disk2': Device error

Failed.

Weird, why the kext that read the NVRAM showed into this KP?

You should test another blade before anything.

 

[HunPro]

I will, but first I try the disabled hibernation sleep. I can put up with some quirks

UPDATE: Did a quick sleep / wake, came back just fine. Hoping that hibernation is the problem, I can live without it.

Update: disabling hibernation solved the issue. It now sleeps solid and wakes up in an instant after a day of sleep.

 

[0134168]

Update: disabling hibernation solved the issue. It now sleeps solid and wakes up in an instant after a day of sleep.

Curious..