On a different subject, I've been seeing low write performance

Question

On a different subject, I've been seeing low write performance

on a laptop's secondary mechanical hard drive. This drive is a 750GB WD Black Scorpio drive, connected via SATA. The write performance I'm seeing on it is around 1MB/s on various loads.

It's partitioned as NTFS storage for Genshin Impact on Windows on one part, where I've been noticing this slow performance for a while now. This suspicion includes even read performance, and has appeared since its Fontaine region appeared in BGM not being played back properly.

The other part is a ZFS pool that's allocated to Linux, as well as a swap area. The ZFS pool is only used sporadically, while the swap area is enabled by default.

Regardless, the IO performance on this drive for both read and write seems to be affected. Should I be concerned for its ability to retain the data that's currently on there? I do have backups, but those don't include everything that's on this drive. I also have plans to exchange this for a SATA SSD, but finances are tight at the moment. That being said, suggestions for good (and affordable) SATA SSD's are welcome. Other than that, should I be acutely concerned about the mechanical hard drive in question?

0

19.10.2023

13 ответов

23 просмотра

Vim Автор вопроса

Martin Rys
I'd check smart. If it's mostly full it could also...

I’ll check up on that yeah. On ZFS fragmentation is more or less inevitable (and unsolvable, see efforts on “block pointer rewrite”), but NTFS shouldn’t be affected by that. Either way, this performance issue being visible on both filesystems is starting to concern me.

0

19.10.2023

Martin Rys

Vim
I’ll check up on that yeah. On ZFS fragmentation i...

Lol NTFS *ABSOLUTELY* has fragmentation issues

0

19.10.2023

Vim Автор вопроса

Martin Rys
I'd check smart. If it's mostly full it could also...

So uh.. I checked up on both of these properties, and I think I found the problem. Time to burden my wallet with a 1TB SSD purchase I guess. Thank God that they went down in price massively by now. Apparently a 1TB MX500 only costs like €50 at this point.

0

19.10.2023

Vim Автор вопроса

Martin Rys
I'd check smart. If it's mostly full it could also...

Стикер

0

19.10.2023

Vim Автор вопроса

Martin Rys
I'd check smart. If it's mostly full it could also...

Стикер

0

19.10.2023

Martin Rys

Vim
So uh.. I checked up on both of these properties, ...

Didn't you say you have an M.2 slot? Oh right but not NVMe?

0

19.10.2023

Martin Rys

Reallocated sectors on an SSD are somewhat expected. Reallocated sectors on a HDD... I think it do be bad.

0

19.10.2023

Vim Автор вопроса

Martin Rys
Didn't you say you have an M.2 slot? Oh right but ...

There's 2 slots, one M.2 slot that can do NVMe, and another that's just regular old SATA ¯\_(ツ)_/¯

0

19.10.2023

Martin Rys

Vim
There's 2 slots, one M.2 slot that can do NVMe, an...

You already have a good NVMe drive? They're pretty much the price of the SATA ones nowadays.

0

19.10.2023

Vim Автор вопроса

Martin Rys
Reallocated sectors on an SSD are somewhat expecte...

I agree, and there is an event that might explain the failure too. Recently I had my laptop in my backpack while it was turned off, and fell on top of said backpack. Another laptop in there actually took a big hit on its display chassis as well, I had to disassemble that one and hammer it back into shape so that it wouldn't push on the LCD anymore. As for the hard drive in this laptop, I guess that also took a hit, even while spun down? Yet another reason to retire these spinning drives in laptops I suppose.

0

19.10.2023

Martin Rys

Yeah, it could have died from the shock. Usually it's a problem when it's in-use, but if it was bad enough you had to hammer chasis back into place...

0

19.10.2023

Vim Автор вопроса

Martin Rys
Yeah, it could have died from the shock. Usually i...

Could have well been the case yeah... Honestly I'm surprised that the amount of damage done to the exterior, didn't mess things up more internally. I guess that's the power of a metal chassis for ya. No elastic deformation whatsoever, but so much more resilient than plastic. Still though, especially in an area where the chassis pierced into the LCD assembly.. I'm surprised that it only left minor permanent damage. I think the crystals are a bit deformed inside of there, but not by much. There are no dead zones anywhere. Considering how thin and fragile a bare LCD panel is (about 1mm), I consider that quite lucky. I do wonder what could've made the drive fail like this though. It's still usable, albeit very slow and I think I'll want to check for data corruption — especially in ZFS. But with the head being parked, and the platters not spinning.. what could still fail in there? Maybe the bearings moved a little from the shock, making the platters ever so slightly rise up on one side? I mean, the tolerances for head-to-platter distance are extremely small, and drives with higher data density make the problem worse (especially SMR drives might illustrate that well). The tinier the speck that the head stores the bit on, the closer it has to be to the platter. Maybe that slight misalignment is causing some scratches on the platter as I'm continuing to use it? No idea to be honest.. and disassembling a spinning drive is a one-way street I guess.

0

19.10.2023

Martin Rys · Accepted Answer

Martin Rys

I'd check smart. If it's mostly full it could also be fragmented to hell, but 1MB is BAD

0

19.10.2023

Похожие чаты

On a different subject, I've been seeing low write performance

13 ответов

Похожие вопросы