Micron 5200 firmware bug

[EffrafaxOfWug]

Unknown-they are in a storage pool with low demands, most of it is used for backup space.

Assuming a RAID pool of some form that still does regular resync/patrol/scrub reads though...? If the performance degradation was as bad as they say in the reddit thread, an array with more than a couple of the 5200 drives >65khr runtime would be unusably slow, if not falling over completely.

Similarly, if the 5100s were affected I'd have hoped we'd have heard about it long before now...

 

[luckylinux]

The 5300 is still supported by Micron so I am watching their download page anxiously (though I have a few years before that's a problem)

No Updates IIRC, I just checked

 

[gregsachs]

Assuming a RAID pool of some form that still does regular resync/patrol/scrub reads though...? If the performance degradation was as bad as they say in the reddit thread, an array with more than a couple of the 5200 drives >65khr runtime would be unusably slow, if not falling over completely.

Similarly, if the 5100s were affected I'd have hoped we'd have heard about it long before now...

Its a storage space with 2x 5100s and 2x HP OEM PM1643, doing simple mirror.
Certainly is not unusuable

 

[nasbdh9]

5100 to 5400 series all use the same main control chip, so theoretically this problem will occur in all of them.

 

[EffrafaxOfWug]

5100 to 5400 series all use the same main control chip, so theoretically this problem will occur in all of them.

I think the MX300 used the same Marvell controller as well, and as already established I've got one of those that's already rolled over 65k hours (and seemingly started again from 0) and is behaving just fine (but a different strand of firmware no doubt, so not directly comparable).

All eyes on other OEMs I guess since it seems like Micron's keeping schtum on this.

 

[tinfoil3d]

Updated fw on 5200 ECO today no problem, the problem initially was i downloadde wrong msecli run file, that was 280mb and wanted rpm. Then i found the correct one and used that, https://www.micron.com/content/dam/...ata-center/storage-executive/msecli-linux.run
as

msecli -F -U D1MU840.bin -n /dev/sd_your_device

Didn't even reset in host, was immediately usable. Noice.

 

[DanAnd]

For those wondering what else might be there ...

root@rescue ~ # ls -la /root/.oldroot/nfs/firmware_update/*/
/root/.oldroot/nfs/firmware_update/adaptec/:
total 36K
drwxr-xr-x  6 root root 4.0K Jul 17  2024 .
drwxr-xr-x 27 root root 4.0K Apr 27 09:31 ..

[...]

drwxr-xr-x  2 root root 4.0K Sep  9  2024 WUH721816ALE6L0
drwxr-xr-x  2 root root 4.0K Sep  8  2025 WUH722016CLE6L1
drwxr-xr-x  2 root root 4.0K Aug  8  2025 WUH722222ALE6L0
drwxr-xr-x  2 root root 4.0K Aug  8  2025 WUH722222ALE6L1
drwxr-xr-x  2 root root 4.0K Sep  5  2024 WUH722422ALE6L1

EDIT 1: all in all it's almost 5GB

root@rescue ~ # du -sh --apparent-size /root/.oldroot/nfs/firmware_update/
4.5G    /root/.oldroot/nfs/firmware_update/

Strangely enough when transferring with scp I only pulled 2.5GB ...

I'm currently trying with rsync over SSH now.

EDIT 2: ignore the 2.5GB Issue listed above. I just needed to run du -sh --apparent-size on the LOCAL copy then it also returns 4.5GB. All is well .

Oh dear! I think there are several interesting firmware versions in that rescue system for my homelab. I really have to double check all my drives and especially samsung oem ssds.

5 of my Micron drives have the problematic firmware as well. Thanks for digging in.

Dan

 

[crlt]

I was hoping this thread was about some old issue that got long resolved... this is ridiculous. I have a few 5200MAX (MTFDDAK1T6TDN) Cisco branded variants that are 1.6TB at 55k hours which are not in the firmware.properties of the Hetzner package so pretty sure I can't use those. I originally got the latest firmware from Cisco but it seems they haven't released anything with the new 5200 files yet.

I also have a 3.84TB branded Cisco drive and one OEM 1.92TB, at least I can save one out of the bunch... guess who is going on a PR campaign if I don't see updated firmware for these?

Cisco got hit too, there are updated firmwares behind paywall:

Micron 5100: D0MU077
Micron 5300: D3CN001

(There could be even newer versions but Micron homepage has nothing)

So basically, if you have a Micron 5x00 and firmware older than end of 2025 or 2026, this bug will affect you after 65535 hours.

There are various Cisco servers models that use these SSDs but I'm not seeing anything in those server firmware package regarding newer firmware. I'm not hopeful...

Edit: It seems like HPE and Lenovo released updated firmware and the EOS for some of these Cisco servers is a few years away so maybe I should be more hopeful.

 

[EffrafaxOfWug]

Edit: It seems like HPE and Lenovo released updated firmware and the EOS for some of these Cisco servers is a few years away so maybe I should be more hopeful.

If you've got access to official support it's probably worth telling them you've heard of this bug, and that HPE and Lenovo have both published patches already, and ask what their plans are for doing the same.

I notice there's been a few more posts on the reddit thread in the last couple of weeks; aside from a few more dozen drives saved from the scrapheap:

• Fujitsu support seemingly also don't care, but their drives use vendor firmware and not custom, so wallcebrf's heroic firmware fishing will be the go-to option
• Micron still keeping shtum on the fact a firmware fix exists
• Another person saying the 5300 series are also affected. Does anyone have any concrete info on this? My oldest 5300's are at about the 50k hour mark so still a fair way to go, but given Micron seemingly didn't clock the bug in the 5200s until late 2025/early 2026, and the last FW for the 5300s was from 2019, I'm not holding out any hope the D3MU[048]01 FW fixes the issue if it does exist on the 5300s

I really, really wish someone with clout would do a name'n'shame on this issue. Micron are being complete cockwombles here.

 

[crlt]

If you've got access to official support it's probably worth telling them you've heard of this bug, and that HPE and Lenovo have both published patches already, and ask what their plans are for doing the same.

I notice there's been a few more posts on the reddit thread in the last couple of weeks; aside from a few more dozen drives saved from the scrapheap:

• Fujitsu support seemingly also don't care, but their drives use vendor firmware and not custom, so wallcebrf's heroic firmware fishing will be the go-to option
• Micron still keeping shtum on the fact a firmware fix exists
• Another person saying the 5300 series are also affected. Does anyone have any concrete info on this? My oldest 5300's are at about the 50k hour mark so still a fair way to go, but given Micron seemingly didn't clock the bug in the 5200s until late 2025/early 2026, and the last FW for the 5300s was from 2019, I'm not holding out any hope the D3MU[048]01 FW fixes the issue if it does exist on the 5300s

I really, really wish someone with clout would do a name'n'shame on this issue. Micron are being complete cockwombles here.

I would if I had access to official support, and people with clout aren't willing to do the name'n'shame because then they wont get fancy micron samples.

 

[Stephan]

Some comments. After upgrade, a few SMART attributes are removed or moved:

 Device Model:     Micron_5300_MTFDDAK960TDT
-Firmware Version: D3MU001
+Firmware Version: D3MU005
 User Capacity:    960.197.124.096 bytes [960 GB]

-196 Reallocated_Event_Count 0x0032   100   100   000    Old_age   Always       -       0
+201 Unknown_SSD_Attribute   0x0032   100   100   000    Old_age   Always       -       0
+249 Unknown_Attribute       0x0022   100   100   000    Old_age   Always       -       1
+250 Read_Error_Retry_Rate   0x0032   100   100   000    Old_age   Always       -       0

For 5100 there is D0MU078.

If your Micron 5x00 drive goes into read-only mode, try to measure PLP buffer caps (LCR meter). If suspicious, solder new ones. Worth a try at current prices.

 

[luckylinux]

For 5100 there is D0MU078.

Is that Cisco / OEM only ? Or where is D0MU078 for the "stock" (non-OEM) Micron 5100 ?

 

[TrevorH]

You might try running /usr/bin/update-smart-drivedb to see if an updated data file might id those unknown attributes.