mkfs.ext4 unbelievably slow formatting 1TB partition

Issues related to applications and software problems
LMHmedchem
Posts: 53
Joined: 2011/03/22 17:32:25
Location: Boston

Re: [solved] mkfs.ext4 unbelievably slow formatting 1TB partition

Post by LMHmedchem » 2020/02/25 18:08:35

hunter86_bg wrote:
2020/02/19 18:41:47
For me the performance is 50% lower than the other ssd.
You should expect better performance.

Move all data away , then create 1 single partition (use all defaults).
I need at least 3 partitions on this drive. How should I create the paritions? Should I use GParted or should I use fdisk? For now I will create 1 partition and test it.

I think that GParted just hacks up an fdisk command but I don't know the set of options that it uses by default. The first partition on the drive is ntfs and needs to be approximately 250GB. Can I use GParted to create an ~250GB ntfs partition that starts at the beginning of the drive?

How do I determine the exact size in MB to use for an ~250GB partition that will be properly aligned (be a multiple of 4096)? I presume that starting it at the beginning will help but what should the last allocated block be.
hunter86_bg wrote:
2020/02/19 18:41:47
Your worst enemy is the disk allignment. If you need to leave a gap - it should be divisible by 4096.
There are 2MB of unallocated space before the first partition. I don't know if that is the correct amount for the first partition to start on a sector boundary. I get different answers with 512 and 4096 sector sizes. If I know where to start and end the first partition to have it properly aligned, then I think I can go from there.

The output of hdparam seems to indicate that the physical sector size is 512 where I would expect it to be 4096.

Code: Select all

sudo hdparm -I /dev/sdb

/dev/sdb:
ATA device, with non-removable media
	Model Number:       Samsung SSD 860 EVO 2TB                 
Configuration:
	Logical		max	current
	cylinders	16383	16383
	heads		16	16
	sectors/track	63	63
	--
	CHS current addressable sectors:   16514064
	LBA    user addressable sectors:  268435455
	LBA48  user addressable sectors: 3907029168
	Logical  Sector size:                   512 bytes
	Physical Sector size:                   512 bytes
	Logical Sector-0 offset:                  0 bytes
	device size with M = 1024*1024:     1907729 MBytes
	device size with M = 1000*1000:     2000398 MBytes (2000 GB)
	cache/buffer size  = unknown
	Form Factor: 2.5 inch
	Nominal Media Rotation Rate: Solid State Device
Is this correct? If so, that doesn't make sense.

LMHmedchem

hunter86_bg
Posts: 2019
Joined: 2015/02/17 15:14:33
Location: Bulgaria
Contact:

Re: mkfs.ext4 unbelievably slow formatting 1TB partition

Post by hunter86_bg » 2020/03/05 15:04:29

Why I mentioned the alingment?!?
Your first partition ended at - 511991549.
Next one started at 511995904.

511995904 - 511991549 is 4355 and it shouldn't be like that, as everything that follows won't be aligned.

For partitioning - anything modern (bad example is fdisk from CentOS 7.0) is enough. I prefer to use fdisk & gdisk but you can use parted, cfdisk (tui based fdisk) , etc.

For the sizing use this example:

Code: Select all

 fdisk /dev/disk
 n
 Primary
 Select default  start
 For end  -> +250G
 Set  type
 
 Repeat using  +<Size>K/M/G for the end  boundary
 
 
Just test with a large partition - if speed is low - RMA the ssd.
Otherwise, if speed is OK - then it was an alignment.

User avatar
TrevorH
Site Admin
Posts: 33219
Joined: 2009/09/24 10:40:56
Location: Brighton, UK

Re: mkfs.ext4 unbelievably slow formatting 1TB partition

Post by TrevorH » 2020/03/05 16:31:32

It's the start sector of the partition that matters, not where it ends. And 511995904 is evenly divisible by 2048 (1MB) so should be fine.
The future appears to be RHEL or Debian. I think I'm going Debian.
Info for USB installs on http://wiki.centos.org/HowTos/InstallFromUSBkey
CentOS 5 and 6 are deadest, do not use them.
Use the FAQ Luke

Post Reply