PSA: SM863 "Endurance Management" firmware feature causing degraded performance over time

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
zac1

zac1

Well-Known Member
Oct 1, 2022
432
358
63
Just a heads up, I got a deal on 20x SM863 1.92TB drives that were in decent shape (no bad sectors, ~2.5PBW) and found that performance was intermittently degraded. As it turns out, there was a recent firmware update released by Dell that removed the "Endurance Management" feature.
Fixes & Enhancements
Fixes:
- Firmware fix to remove Endurance Management feature, in order to maintain drive performance through life of drive.
Version
GB57, A00
Release date
30 Sep 2022
Download Type
Firmware
Click to expand...
A fellow STH'er (EDIT: @mmk) clued me in:
Also have some SM863s at both home and work. They have an issue where some endurance management screws up and performance drops to something like some ten MB/s....at least if the drives have been heavily used.

Dell put out new firmware to fix that in August this year..even though these drives are rather old at this point.

Your issue sounds _a lot_ like what i've been seeing. From installing the updated firmware it seems like the sooner you do it the better. If the drive is already exhibiting severe issues the it may already be too late. This is based on some perhaps 30 updated drives behind raid controllers...it may well be that there is just one or two drives causing the issues there and the rest are fine though.

Also something to note is the drives i have at home have had an easy life and work just fine.
Click to expand...
The problem is the firmware is not available/impossible to find for non-Dell OEM drives. The workaround is to truncate the Dell firmware to just under 1MB and force it on the drive using hdparm.

Potential problems: the drive can be bricked, the version string reported by SMART does not update, and some SMART values are reset.

And to think I almost bought 50 more...

Huge thank you to @mmk for the lead on this. I might've sunken a few thousand dollars into poorly performing drives had they not taken their time to clue me in to the potential performance issues! And thank you to @Stephan, of course, for stressing the importance of choosing drives with freely available firmware updates — I get it now! ;)
 
Last edited:
  • Like
Reactions: itronin, Marsh, mach3.2 and 2 others
M

MPServers

New Member
Feb 4, 2024
23
16
3
zac1 said:
...
The problem is the firmware is not available/impossible to find for non-Dell OEM drives. The workaround is to truncate the Dell firmware to just under 1MB and force it on the drive using hdparm.

Potential problems: the drive can be bricked, the version string reported by SMART does not update, and some SMART values are reset.
...
Click to expand...
Reviving an old thread, but...

So, I recently pulled 24 of these (Samsung SM863 1.92 TB drives) from a system we retired and thought these would make some good "play" drives in some servers I have in my home lab.

I was putting them all into some PowerEdge R730's and doing Diskmark tests and noticed some TERRIBLE speeds, narrowing it down to one drive in particular with just abysmal *read* speeds (single digit MB/s).

After reading all of these threads, I thought, sure, I can grab the Dell update, do the mods, and flash it. Well, I can get it to flash and it reports "done" without any errors, but even after power cycling the drives, it's still reporting the firmware version as GXM1103Q instead of (what I expected) GXM1203Q.

Maybe that's because this is Dell firmware and it would try to set the firmware version to GB57 or whatever so, like you said, the version report isn't correct... But is that really true in general? I saw your comment to that effect but I didn't remember seeing anyone else reporting it.

Meanwhile, that drive was still terribly slow so maybe it's just too late to save that one. I did a secure erase as well and I'm about to retest but I don't have much hope. I'll just throw that into a bucket of odd parts and see if there's anything in the future that would help it out.
 
J

joerambo

New Member
Aug 30, 2023
10
0
1
Also a bit of highjack of a thread, but maybe someone already has:

PM893 2.5" : JXTC404Q firmware

Need to update some drives, as always Samsung and resellers are very "helpful".
 
M

MPServers

New Member
Feb 4, 2024
23
16
3
I had a bit of an epiphany last night when I was doing more testing. I finally just gave up and swapped out that "bad" drive for another fresh one, and to my surprise, it had the same issue. Well, that got me thinking maybe (probably) it's either the controller or that specific drive bay (Dell R730... bay 4). I tabled the thought until this morning when I went back out there, blew out everything, reseated the drive bay cabling along with some deoxit as well. Lo and behold, the drive performance shot back up to normal.

So it turns out I was chasing my tail the whole time on the firmware issue when it was really just a poorly seated cable or something. I thought about trying one thing at a time (deoxit on the drive bay, then working back by doing that on the cabling to the motherboard as well) but I was in a hurry to get that all done before the work day started.

It's weird that what I assume was some poor contact somewhere was resulting in slow *read* speeds, but normal *write* speeds. That seems unusual. But the good news is, I created a 7-drive software stripe (disk 0 is the OS... this R730 doesn't have a raid controller) and it all seems happy and normal now.

But I would definitely recommend that if you're seeing any similar issues, while it may be the firmware, it may also be some bad cabling, dirty sockets, or just need the old unplug/replug trick.

I'm still glad I went through the process of doing this "back door" firmware updating because it was fun to try things out anyway, and as long as I learn something new, that's fine with me. It wasn't a waste of time. :)
 
  • Like
Reactions: zac1 and pimposh
You must log in or register to reply here.
Top Bottom