With old MBR format, you can copy (cp or dd), then resize the last partition, and then resize file system in it.
Wth new GPT format, you can't do that, because GPT has secondary partition table at the end.  So, you need to use tools like Clonezilla.

I have an old Dell XPS 9560 notebook.  It has a dying SSD.

The SSD is M.2 SATA, which is not a good choice for performance.
The XPS can support NVMe drives in the same socket, so I went shopping for 
one.

(I bought this computer used, before the pandemic.  The seller said he had 
problems with the original SSD so he replaced it with this SATA drive.  
There is a chance I'll have trouble with an NVMe drive.)

I checked: this XPS does not support HMB.  HMB has been an important 
technology for making NVMe drives cheaper.  It allows some of the host 
computer's RAM to be used by the drive, eliminating the need for on-SSD 
RAM.

So: I kind of need a drive with on-board RAM.
There are not that many of them left.

I also prefer TLC as opposed to QLC since TLC generally has better 
endurance.  And sometimes higher speed, especially for sustained writes.

I found it hard to search for an SSD with these attributes.  They aren't 
often spelled out in listings so Amazon search and AliExpress search are 
not great.

I ended up buying one of these: 
<https://www.aliexpress.com/item/1005003788387057.html>
That comes in several size variants, but I don't want the -T models since 
they have no on-board DRAM.

I will find out if the "metal heat dissapation vest" prevents it fitting 
in my notebook.  If it doesn't fit, I can find it a home elsewhere.

Mystery:

In the past DRAMs on NVMe drives seemed to be 1G (small sample of 
inexpensive drives).

HMB buffers seem to be much smaller.  For example, on my desktop, the 
Crucial P3 Plus 1TB NVMe's HMB seems to be 32MiB.  That seems to be so 
much smaller that it must be doing something less than the on-board DRAM 
would do.

Finding HMB size:

On my desktop, I followed some AI advice from DuckDuckGo:

 $ sudo nvme get-feature /dev/nvme0n1p1 -H -f 0x0d

The last line of output said:
	Host Memory Buffer Size                  (HSIZE): 8192

The AI advice said that that was the the size of the HMB in bytes. That 
turns out not to be the case.  It is the size in memory pages (4k on my 
machine, I think).  This is from a note to an answer in an old 
stackexchange page:

<https://unix.stackexchange.com/questions/681131/how-to-check-change-nvme-hmb-on-linux>

So AI helped but was wrong in an important detail (by a factor of 4k).  
The stackexchange answer was similarly wrong but had a correcting note 
(coincidence???).

There may be future war stories when I get this drive and try to install 
it.

Any recommendations for cloning the old drive, skipping bad spots?

- I have Windows and Linux partitions

- I have an NVMe to USB device.

- I can boot the computer from a USB stick so I can copy without
  either source or destination partitions being mounted.

- the new drive is larger than the old one.
------------------------------------
Description: GTALUG Talk
Unsubscribe via Talk-unsubscribe@lists.gtalug.org
Start a new thread: talk@lists.gtalug.org
This message archived at https://lists.gtalug.org/archives/list/talk@lists.gtalug.org/message/QG7II62GZ2D27JIXENFUF42EG7XNMENJ/