Looking for advice on 4TB NVME SSDs for workstation use.

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

Analogue Blacksheep

New Member
Sep 8, 2019
9
4
3
Hello all, I was wondering if I could get some suggestions/recommendations on some 4tb NVME SSDs. I'm a graphic designer by trade and plan on using my computer for profession/business use. I often deal with very large file sizes and intensive workloads, so I need something that is realible and does the job.

Why 4tb you may ask? I'm looking for something that works in the long term and at some point I plan on getting two 4tb hard drives that I can use as an external backup so if I can mirror the drives exactly that would be nice. I know the Sabrent Rocket 4TB M.2 exists, but I want to know if there is anything in the enterprise space that would be worth my time. I'm especially interested in the drive having Power Loss Protection (PLP) as well. My build is currently in a Lone L5, an ultra small SFF build that has been serving me well. The downside is however, I can only use 7 inch 2.5 drives or M.2 based solutions. If I end up with a 22110 drive I plan on taking an ADT-Link R44SF adapter an modifying it via 3D printing.

One drive I have considered is the Samsung PM983 3.84TB Enterprise M.2. It's got PLP and seems good on paper. Another drive I've looked at is the Western Digital WD Ultrastar DC SN640 M.2 3.84TB, but it is currently unavailable as of writing.

If anyone has any suggestions that would be great! Thanks!
 

TLN

Active Member
Feb 26, 2016
523
84
28
34
I got very good experience with PM1725 drives. Got pair of 3.2Tb, currently have 6.4 and 3.2Tb drives. They're PCIe 3.0 8x, compared to M.2 which is 4x only.
Drives rated to crazy number of data written and got all PLP features.
Here's comparsion between multiple drives:
Imgur
FusionIO ioDrive2 1.2Tb
Intel P3605 1.6Tb
Samsung PM1725 3.2Tb.

All was done via Esxi hypervisor, with few more VM running.
With recent 3.2Tb drive I got 6000/3000MBPS throughput.

Just realized you got SFF case.You can score PM1725 in 2.5" format on ebay for $400 or so for 3.2Tb model. That actually one of the reasons I'm going with Cerberus case and mATX mobo over ITX.
 
  • Like
Reactions: Analogue Blacksheep

TLN

Active Member
Feb 26, 2016
523
84
28
34
@TLN - Thanks for the suggestion. I'm going to keep an eye on eBay in the case anything comes up.

---

I also came across this on SCAN for a decent price, the Samsung PM983 3.84TB Enterprise NF1 PCIe NVMe SSD. Delock make an adapter, the 64077, so it could fit in my case. Only question is, would it work on Windows 10 Pro?
Yep, it should work.
There's a person selling 7.68 NVMEs here: https://forums.servethehome.com/ind...p-dl325-epyc-24-core-7402p.27907/#post-257533

Btw, for 2.5 Devices and M.2 slot I highly recommend intel M.2 to U.2 adapter:
https://i.imgur.com/WBhQ5Pg.jpg
 

Dreece

Active Member
Jan 22, 2019
503
160
43
I have the PM1725a's, and Intel P4510's.... the Samsungs are far superior at low queue-depth work, which is what you ideally need because of your workload. If you can get a good price on PM1725a, get that, ideally a 6.4tb card, and never look back, 3.2tb vanishes quick, trust me.

I'd avoid the 983, I've had them too and their QD1 performance ain't that great, they're better suited to cheap datacentre cloud-storage, not really for high-end workstation use, probably find them slower than most of today's consumer level NVME devices. Though to be fair all NVME drives will be more than adequate for your particular workload.

If you want great workstation performance, longevity, acceptable power-usage, reasonably small in size, no flaffing around with U.2 adapter cards with or without cables etc... the PM1725a is my recommendation... if 6.4tb is out of budget, hit the 3.2tb and just keep an eye out for a 6.4.tb version when one appears at a price which suits you and then just buy that and sell off the 3.2tb one.
 

Analogue Blacksheep

New Member
Sep 8, 2019
9
4
3
@Dreece - Thanks for the suggestion, however unfortunately my case can only fit 7mm thick drives or adapters which is proving to be a major headache and I don't have any spare PCIE slots.

I've also been looking at the WD Gold U.2's, the WD Ultrastar's SN640 and the Micron 7300 (Although I've noticed the latter seems to be missing PLP). I am still open to the idea of other form factors via an adapter like 22110 M.2 and NF1. Been also keeping an eye on eBay a little more as well.

If it helps, I'm currently using an Adata XPG SX8200 Pro 1TB as my current boot drive.
 

Dreece

Active Member
Jan 22, 2019
503
160
43
...I don't have any spare PCIE slots.
Honestly looking forward to the day they throw that limitation in the bin and give us some kind of multiplexer port so we can have as many pcie lanes as we want though they already do that somewhat with plx bridges. Ideal world we get a multiplexing oculink connection and we get the option to house all our cards anywhere we want... mind you thunderbolt was supposed to be that solution in a way.

I guess if pcie cards are off the table then you've got quite a lot to choose from. Well do let us know which one you end up going for. Far too many come here, drop a post, and then vanish, would be nice to hear what they ended up doing for a change.
 
  • Like
Reactions: Analogue Blacksheep

Analogue Blacksheep

New Member
Sep 8, 2019
9
4
3
@Dreece - Agreed, I'm surprised OCuLink isn't used more often outside of the enterprise space. I would trade the four SATA ports on my board for two of them. I'm mostly in Small Form Factor PC circles and if someone outside of Asrock Rack used them for an ITX board I could see them being a game changer that space.

Anyway, I'll let you guys know what I end up with. It might be a while before I show something, but I'll let you know. I'm sure there is someone else on the internet asking the same question. It would be a little silly not to share a potential solution.
 

Analogue Blacksheep

New Member
Sep 8, 2019
9
4
3
Well folks, I've currently narrowed it two options: The Micron 7300 Max 3.2TB or the Micron 7300 Pro 3.8TB. You can sum it up as the former can hold more files and is £173‬ less which fits much better in my budget, while the latter has much better endurance and IOPS. Turns out both have power loss protection in flight, so it's better than nothing.

Just thinking though. If I did go with the 7300 Pro, I could use my Adata XPG SX8200 Pro drive to edit active projects and then move them back to the U.2 for long term internal storage. But the irony of that could be, I might end doing more writes that way! If I had the budget I would go with the 7300 MAX drive 6.4TB and get the best of both worlds, but it's £1,631.99 :eek:. I don't really work with videos or RAW files right now, so I'm fine with 3TB to 4Tb for now. Plus I'm using the Serif Affinity triplets (Photo, Designer, Publisher) for my work and it's way more nippy then the Adobe Creative Cloud.

I've got time to think things over.
 
  • Like
Reactions: Dreece
Sep 4, 2017
47
25
18
46
Given that there is a deadly, complete data loss bug on Samsung PM1725a drives that is fixed by a firmware update which is totally unobtainable by any end user, IMHO it's irresponsible to recommend PM1725a to anyone. They are completely unsupported by Samsung.
 

Dreece

Active Member
Jan 22, 2019
503
160
43
Given that there is a deadly, complete data loss bug on Samsung PM1725a drives that is fixed by a firmware update which is totally unobtainable by any end user, IMHO it's irresponsible to recommend PM1725a to anyone. They are completely unsupported by Samsung.
It is also irresponsible to effectively claim all PM1725a drives have a deadly bug. As far as I was aware the only problem was on Linux platform with an early OEM batch. Unless I completely missed a PM1725a major issue?

I have 6 of these cards running at one site for the past 4 years now, excessively hit hard by a gigantic MSSQL cluster, never missed a heart beat. I have 3 myself, one for video editing, one for databases and one for just being empty.

Also if anyone is running a homelab/serious-computer without a backup, well that is a whole different issue - indeed everything is deadly when one has no DR scheme in execution.

My old dusty laptop with its OCZ Vertex SSD is deadly-incarnate.

PS. one has to also keep in mind that many who come here are keen on purchasing reclaimed gear, and in the majority of these such cases, most things are barely supported by their original manufacturers, that's sometimes a risk one is willing to take due to budget constraints.

In the ideal world it would be more helpful to detail what bugs there are with references, also which batches or oem models the bugs actually affect.
 
Last edited:
Sep 4, 2017
47
25
18
46
Document Display | HPE Support Center

IMPORTANT: This HPK1 firmware upgrade is considered a critical fix and is required to prevent the issue detailed below. HPE strongly recommends immediate application of this critical fix. Neglecting to perform the required action could leave the server in an unstable condition, which could potentially result in SSD unresponsiveness. By disregarding this notification and not performing the recommended resolution, the customer accepts the risk of incurring future related errors.

When an unexpected power-cycle occurs, the power-loss event may not be detected by the PM1725a's firmware, resulting in loss of namespace on subsequent power-on. If the issue already occurred, formatting the drive is the only recovery method available.
There is actually a second cache management bug as well referenced by Dell

Dell Express Flash NVMe PCIe SSD PM1725a firmware release | Driver Details | Dell Canada
 
Sep 4, 2017
47
25
18
46
I have had my fill of unsupported OEM crap after my experience with WD/SanDisk/HGST disowning the SN260 series.
 

LikesChoco

New Member
Apr 23, 2020
2
0
1
Before I found out that this wasn't directly supported by Samsung I picked a PM1725a up.

The firmware it came with is GPNA0A3Q. From the string it sounds like it's probably the original firmware.

VMware requires GPNA5B3Q for support:
VMware Compatibility Guide - vfrc

SuperMicro indicates GPNAAB3Q is available:
FAQ Entry | Online Support | Support - Super Micro Computer, Inc.

OEM versions of the drive have been updated with firmware available for download:
  • HP Enterprise released a recent firmware for their OEM branded MT006400KWHAE:
    HPESC Search Results
  • Dell released a somewhat recent firmware for their OEM branded Dell Express Flash NVMe:
    Challenge Page
Is there anywhere / any way to obtain the latest Samsung firmware for this drive?

From reading this thread it sounds grim.

No luck cross flashing it to an OEM version either? I was considering trying that myself (looks like most of the identifying data is in payload.json and package.xml files) but don't want to risk it if it might be possible to get the Samsung generic firmware.
 

Don.key

Member
Apr 10, 2020
48
22
8
Is there anywhere / any way to obtain the latest Samsung firmware for this drive?

From reading this thread it sounds grim.

No luck cross flashing it to an OEM version either? I was considering trying that myself (looks like most of the identifying data is in payload.json and package.xml files) but don't want to risk it if it might be possible to get the Samsung generic firmware.
I second that question! I have GPNA0A3Q based PM1725a as well and would like to upgrade firmware to version with critical bugfixes.
 

Don.key

Member
Apr 10, 2020
48
22
8
Follow up:

I actually decided to risk one of my MZPLL6T4HMLS-000MV drives and perform flash of HP HPS1 Firmware on it using Linux package. I installed CentOS on the machine, then the rpm provided on the page linked by LikesChoco in his post above.
The actual files that perform update can be found under /usr/lib/x86_64-linux-gnu/firmware-hdd-8e8ddc5265-HPS1-3.1

The key file that does the job is called .setup (yes, with a dot, a hidden file), it reads config from CP040131.xml, here you can find all names of drives this tool will recognize, to fool the flasher you need to replace all mentioning of MT006400KWHAE with whatever your drive name is (mine is MZPLL6T4HMLS-000MV). Then .setup will recognise your drive and present you with a choice if you have more then one (as I do).

But flash failed, logs can be found under /var/cpq/, see below for mine. Drive is OK, I power cycled the machine and it works fine as before.

Code:
[ Apr 30 20:14:34 ] The following 2 devices are supported
[ Apr 30 20:14:34 ]  * Drive ***CENSORED_SERIAL*** (NVME HBA NVMe SSD Controller 172Xa/172Xb in Slot ATTR_VALUE_SLOT_UNKNOWN)
[ Apr 30 20:14:34 ]  * Drive ***CENSORED_SERIAL2*** (NVME HBA NVMe SSD Controller 172Xa/172Xb in Slot ATTR_VALUE_SLOT_UNKNOWN)
[ Apr 30 20:14:34 ]
[ Apr 30 20:14:34 ] Reason:
[ Apr 30 20:14:34 ] Meet all requirements
[ Apr 30 20:14:34 ]
[ Apr 30 20:14:34 ]
[ Apr 30 20:14:34 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrTFWDL
[ Apr 30 20:14:34 ]
[ Apr 30 20:14:34 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrFastWrBufModes
[ Apr 30 20:14:34 ]
[ Apr 30 20:14:34 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrFastWrBufModes
[ Apr 30 20:14:34 ]
[ Apr 30 20:14:34 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrSPCRev
[ Apr 30 20:14:34 ]
[ Apr 30 20:14:34 ] 3526245248: Sending SCSI Inquiry request to device NONSAPD:3246091974
[ Apr 30 20:14:34 ] 3526245248: CDB[00:05] 12 00 00 00 5F 00
[ Apr 30 20:14:34 ] 3526245248: SCSI Inquiry request failed
[ Apr 30 20:14:34 ]     i32LowLevelStatus ... 0x00000000
[ Apr 30 20:14:34 ]     wCommandStatus ...... 0x00000000
[ Apr 30 20:14:34 ]     bScsiStatus ......... 0x00000000
[ Apr 30 20:14:34 ]
[ Apr 30 20:14:34 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrTFWDL
[ Apr 30 20:14:34 ]
[ Apr 30 20:14:34 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrFastWrBufModes
[ Apr 30 20:14:34 ]
[ Apr 30 20:14:34 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrFastWrBufModes
[ Apr 30 20:14:34 ]
[ Apr 30 20:14:34 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrSPCRev
[ Apr 30 20:14:34 ]
[ Apr 30 20:14:34 ] 3526245248: Sending SCSI Inquiry request to device NONSAPD:409887668
[ Apr 30 20:14:34 ] 3526245248: CDB[00:05] 12 00 00 00 5F 00
[ Apr 30 20:14:34 ] 3526245248: SCSI Inquiry request failed
[ Apr 30 20:14:34 ]     i32LowLevelStatus ... 0x00000000
[ Apr 30 20:14:34 ]     wCommandStatus ...... 0x00000000
[ Apr 30 20:14:34 ]     bScsiStatus ......... 0x00000000
[ Apr 30 20:14:35 ]
[ Apr 30 20:14:35 ] 3491456768: Sending Pause Background Activity command to device HBA:3692555625
[ Apr 30 20:14:35 ] Device HBA:3692555625 does not support OPERATION_WRITE_BMIC_COMMAND
[ Apr 30 20:14:35 ]
[ Apr 30 20:14:35 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrFFWDL
[ Apr 30 20:14:35 ]
[ Apr 30 20:14:35 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrTFWDL
[ Apr 30 20:14:35 ]
[ Apr 30 20:14:35 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrVendor
[ Apr 30 20:14:35 ]
[ Apr 30 20:14:35 ] 3491456768: Sending SCSI Inquiry request to device NONSAPD:3246091974
[ Apr 30 20:14:35 ] 3491456768: CDB[00:05] 12 00 00 00 5F 00
[ Apr 30 20:14:35 ] 3491456768: SCSI Inquiry request failed
[ Apr 30 20:14:35 ]     i32LowLevelStatus ... 0x00000000
[ Apr 30 20:14:35 ]     wCommandStatus ...... 0x00000000
[ Apr 30 20:14:35 ]     bScsiStatus ......... 0x00000000
[ Apr 30 20:14:35 ]
[ Apr 30 20:14:35 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrSPCRev
[ Apr 30 20:14:35 ]
[ Apr 30 20:14:35 ] 3491456768: Sending SCSI Inquiry request to device NONSAPD:3246091974
[ Apr 30 20:14:35 ] 3491456768: CDB[00:05] 12 00 00 00 5F 00
[ Apr 30 20:14:35 ] 3491456768: SCSI Inquiry request failed
[ Apr 30 20:14:35 ]     i32LowLevelStatus ... 0x00000000
[ Apr 30 20:14:35 ]     wCommandStatus ...... 0x00000000
[ Apr 30 20:14:35 ]     bScsiStatus ......... 0x00000000
[ Apr 30 20:14:35 ]
[ Apr 30 20:14:35 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrDevBufSize
[ Apr 30 20:14:35 ]
[ Apr 30 20:14:35 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrFastWrBufModes
[ Apr 30 20:14:35 ]
[ Apr 30 20:14:35 ] ===========================================================
[ Apr 30 20:14:35 ] ===========================================================
[ Apr 30 20:14:35 ] Flashing device: Drive ***CENSORED_SERIAL*** (NVME HBA NVMe SSD Controller 172Xa/172Xb in Slot ATTR_VALUE_SLOT_UNKNOWN)
[ Apr 30 20:14:35 ]     ID:               NONSAPD:3246091974
[ Apr 30 20:14:35 ]     WWID:             ***CENSORED_WWID***
[ Apr 30 20:14:35 ]     Model:            SAMSUNG MZPLL6T4HMLS-000MV
[ Apr 30 20:14:35 ]     Type:             ATTR_VALUE_INTERFACE_NVME_SSD
[ Apr 30 20:14:35 ]     Drive Type:
[ Apr 30 20:14:35 ]     Firmware:         GPNA
[ Apr 30 20:14:35 ]     Pending Firmware: N/A
[ Apr 30 20:14:35 ]     Vendor:
[ Apr 30 20:14:35 ]     Immediate:
[ Apr 30 20:14:35 ]     FFWDL/TFWDL:      0/0
[ Apr 30 20:14:35 ]     SPC Revision:
[ Apr 30 20:14:35 ]     ActivateUcode:
[ Apr 30 20:14:35 ]     POA Support:
[ Apr 30 20:14:35 ]     HRA Support:
[ Apr 30 20:14:35 ]     DeviceBufSize:
[ Apr 30 20:14:35 ]     FastWrBufModes:   0x0
[ Apr 30 20:14:35 ]     SuppWrBufModes:   0x13
[ Apr 30 20:14:35 ]     Serial Number:    ***CENSORED_SERIAL***
[ Apr 30 20:14:35 ]     Status:           ATTR_VALUE_STATUS_OK
[ Apr 30 20:14:35 ]     Controller:       NVME HBA NVMe SSD Controller 172Xa/172Xb in Slot ATTR_VALUE_SLOT_UNKNOWN
[ Apr 30 20:14:35 ]     Ctrl Immediate:
[ Apr 30 20:14:35 ]
[ Apr 30 20:14:35 ] ===========================================================
[ Apr 30 20:14:35 ]
[ Apr 30 20:14:35 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrDevBufSize
[ Apr 30 20:14:35 ] NVMe Flash using Native Commands
[ Apr 30 20:14:35 ] NVME Flash is supported DEFERRED update
[ Apr 30 20:14:35 ] Using firmware image ./HPS1.rel
[ Apr 30 20:14:35 ] Start Flashing NVMe Drive...
[ Apr 30 20:14:35 ] Performing the NVME Firmware flash....
[ Apr 30 20:14:39 ] Failed to flash the NVME Drive
[ Apr 30 20:14:39 ] Nvme Firmware flash status : failure
[ Apr 30 20:14:39 ] Apr 30 20:14:39 Complete flash, status 107
[ Apr 30 20:14:39 ] Flash failed
[ Apr 30 20:14:39 ]
[ Apr 30 20:14:39 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrFFWDL
[ Apr 30 20:14:39 ]
[ Apr 30 20:14:39 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrTFWDL
[ Apr 30 20:14:39 ]
[ Apr 30 20:14:39 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrVendor
[ Apr 30 20:14:39 ]
[ Apr 30 20:14:39 ] 3491456768: Sending SCSI Inquiry request to device NONSAPD:3246091974
[ Apr 30 20:14:39 ] 3491456768: CDB[00:05] 12 00 00 00 5F 00
[ Apr 30 20:14:39 ] 3491456768: SCSI Inquiry request failed
[ Apr 30 20:14:39 ]     i32LowLevelStatus ... 0x00000000
[ Apr 30 20:14:39 ]     wCommandStatus ...... 0x00000000
[ Apr 30 20:14:39 ]     bScsiStatus ......... 0x00000000
[ Apr 30 20:14:39 ]
[ Apr 30 20:14:39 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrSPCRev
[ Apr 30 20:14:39 ]
[ Apr 30 20:14:39 ] 3491456768: Sending SCSI Inquiry request to device NONSAPD:3246091974
[ Apr 30 20:14:39 ] 3491456768: CDB[00:05] 12 00 00 00 5F 00
[ Apr 30 20:14:39 ] 3491456768: SCSI Inquiry request failed
[ Apr 30 20:14:39 ]     i32LowLevelStatus ... 0x00000000
[ Apr 30 20:14:39 ]     wCommandStatus ...... 0x00000000
[ Apr 30 20:14:39 ]     bScsiStatus ......... 0x00000000
[ Apr 30 20:14:39 ]
[ Apr 30 20:14:39 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrDevBufSize
[ Apr 30 20:14:39 ]
[ Apr 30 20:14:39 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrFastWrBufModes
[ Apr 30 20:14:39 ]
[ Apr 30 20:14:39 ] Fetching external attribute ::hal::STORAGE_API::PhysicalDriveAttrSuppWrBufModes
[ Apr 30 20:14:39 ]
[ Apr 30 20:14:39 ] ===========================================================
[ Apr 30 20:14:39 ] Completed flash of device: Drive ***CENSORED_SERIAL*** (NVME HBA NVMe SSD Controller 172Xa/172Xb in Slot ATTR_VALUE_SLOT_UNKNOWN)
[ Apr 30 20:14:39 ]     ID:               NONSAPD:3246091974
[ Apr 30 20:14:39 ]     WWID:             ***CENSORED_WWID***
[ Apr 30 20:14:39 ]     Model:            SAMSUNG MZPLL6T4HMLS-000MV
[ Apr 30 20:14:39 ]     Type:             ATTR_VALUE_INTERFACE_NVME_SSD
[ Apr 30 20:14:39 ]     Drive Type:
[ Apr 30 20:14:39 ]     Firmware:         GPNA
[ Apr 30 20:14:39 ]     Pending Firmware: N/A
[ Apr 30 20:14:39 ]     Vendor:
[ Apr 30 20:14:39 ]     Immediate:
[ Apr 30 20:14:39 ]     FFWDL/TFWDL:      0/0
[ Apr 30 20:14:39 ]     SPC Revision:
[ Apr 30 20:14:39 ]     ActivateUcode:
[ Apr 30 20:14:39 ]     POA Support:
[ Apr 30 20:14:39 ]     HRA Support:
[ Apr 30 20:14:39 ]     DeviceBufSize:
[ Apr 30 20:14:39 ]     FastWrBufModes:   0x0
[ Apr 30 20:14:39 ]     SuppWrBufModes:   0x13
[ Apr 30 20:14:39 ]     Serial Number:    ***CENSORED_SERIAL***
[ Apr 30 20:14:39 ]     Status:           ATTR_VALUE_STATUS_OK
[ Apr 30 20:14:39 ]     Controller:       NVME HBA NVMe SSD Controller 172Xa/172Xb in Slot ATTR_VALUE_SLOT_UNKNOWN
[ Apr 30 20:14:39 ]     Ctrl Immediate:
[ Apr 30 20:14:39 ]  Flash failed
[ Apr 30 20:14:39 ] ===========================================================
[ Apr 30 20:14:39 ] ===========================================================
[ Apr 30 20:14:39 ] Escalating flash error status
[ Apr 30 20:14:39 ]
[ Apr 30 20:14:39 ] 3491456768: Sending Resume Background Activity command to device HBA:3692555625
[ Apr 30 20:14:39 ] Device HBA:3692555625 does not support OPERATION_WRITE_BMIC_COMMAND
[ Apr 30 20:14:39 ] Devices [Drive ***CENSORED_SERIAL*** (NVME HBA NVMe SSD Controller 172Xa/172Xb in Slot ATTR_VALUE_SLOT_UNKNOWN)]: 107
[ Apr 30 20:14:39 ]   Flash failed
[ Apr 30 20:14:39 ]  Internal Exit Status: 107
[ Apr 30 20:14:39 ] See log at /var/cpq/CP040131_2020_04_30_20_14_32.log
[ Apr 30 20:14:39 ] ============ Summary ============
[ Apr 30 20:14:39 ] Smart Component Finished
[ Apr 30 20:14:39 ]
[ Apr 30 20:14:39 ] Summary Messages
[ Apr 30 20:14:39 ] ================
[ Apr 30 20:14:39 ] 1 flash attempt failed
[ Apr 30 20:14:39 ] User opted to not flash 1 device
[ Apr 30 20:14:39 ] See log at /var/cpq/Component.log for details
[ Apr 30 20:14:39 ]
[ Apr 30 20:14:39 ] Exit Status: 7
[ Apr 30 20:14:39 ] Flash failed
[ Apr 30 20:14:39 ] A fatal error occurred. See engineering log for details.
As it stands now those drives are stuck with the original firmware. As far as I could find, the critical bugs for this drive where:

Problems with PLP:

Cache corruption on dell drives:

Cache corruption under Linux with NVME Queue above 64:
This is possible the same issue as above.

I guess as long as your driver / os does not have a NVME Queue depth >64 and you have a stable power to the machine you will be alright.
 
Last edited:

111alan

Active Member
Mar 11, 2019
290
107
43
Haerbing Institution of Technology
It's more than cache corruption. Dell's PM1725a experienced several FW updates. The initial one is 1.0.0, then 1.0.2, 1.0.3, 1.0.4, 1.1.1, and lastly 1.1.2. There's a lot of things they fixed. Dell seems to have removed the update history on its webpage but I saved those files. There are many important fixes like improving the power-loss handling(may stop drive from bricking in some situations), and most updates are at urgent level.

That's why I always recommend Dell drives, or HP drives if you can't get a dell one. PM1725 series SSDs are among the most powerful NAND drives I've ever tested(which are, almost every NVMe enterprise drives from big vendors), but it's hard to find one from a reliable OEM.
PM1725a urgent FW.png
 

Evan

Well-Known Member
Jan 6, 2016
3,346
598
113
Some people say don’t use an OEM like dell or HPE, but so long as you can use a machine to update the disks it’s by far the easiest way compared to some unknown OEM.
 

Don.key

Member
Apr 10, 2020
48
22
8
Usually try to stay away from OEM versions and get manufacturer HW (Intel excels in this) but in the world of NVMe SSDs that formula seems not to work. I am very disappointed with Samsung I must admit....

Regarding the PN1725a: Interestingly HP had not so many updates for those drives, essentially they released only one critical update HPK1:

Firmware corrects an issue where the drive will be in a failed state after an unexpected power loss. When this occurs the drive will not recover after subsequent power cycles and will not be accessible by the system configuration and Host applications.