Ext4 filesystems created without partition tables may have affected my crypto rig

Ask Fedora
1

OS : Fedora 34

One month back, I built an AMD Ryzen based desktop PC which does Chia coin farming.

I have several 12 TB and 18 TB mechanical Hard drives (Western Digital and Seagate) attached to the motherboard. For each disk, I created an ext4 filesystem covering the whole disk. i.e. no LVMs involved.

Typically, in linux, when you want to create an ext4 filesystem, following is what you would do.

Step1. Partition the HDD using fdisk or parted.

 $ parted -s /dev/sdc mklabel gpt

 $ sudo parted -s /dev/sdc unit mib mkpart primary 1 100%

Result of the above commands is a /dev/sdc1 partition which uses GPT partitioning scheme.

Step2. Run mkfs.ext4 command in the newly created /dev/sdc1

mkfs.ext4 /dev/sdc1

Now /dev/sdc1 is ready to be mounted on the directory tree of Linux

But, my friend, who is a java developer and not very familiar with Linux and doing Chia farming was creating the filesystems directly from the attached SATA HDD as shown below. i.e. he was running mkfs.ext4 command directly on the newly attached disks WITHOUT paritioning them first. So, I too did the same.

Following is the output from Fedora 34 when I ran mkfs.ext4 command directly on one of my 14TB HDDs

$ sudo mkfs.ext4 /dev/sdc
mke2fs 1.45.6 (20-Mar-2020)
Creating filesystem with 3418095616 4k blocks and 427261952 inodes
Filesystem UUID: 0cb9a4a8-19b5-4d93-91de-89ecbf31f643
Superblock backups stored on blocks: 
	32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632, 2654208, 
	4096000, 7962624, 11239424, 20480000, 23887872, 71663616, 78675968, 
	102400000, 214990848, 512000000, 550731776, 644972544, 1934917632, 
	2560000000


Allocating group tables: done                            
Writing inode tables: done                            
Creating journal (262144 blocks): done
Writing superblocks and filesystem accounting information: done

When I ran the below command to query what the partition table type of the above disk, it was listed as ‘loop’ . If I partitioned the disk using gpt partitioning scheme, it would have been listed as gpt

$ sudo parted /dev/sdc print
Model: ATA WDC WD141KFGX-68 (scsi)
Disk /dev/sda: 14.0TB
Sector size (logical/physical): 512B/4096B
Partition Table: loop     ----------------------------------> here
Disk Flags: 

Number  Start   End     Size    File system  Name     Flags
 1      1049kB  14.0TB  14.0TB               primary

I was in a rush and I ended up formatting all the HDDs like above. Now, after 1 month, the filesystems created like above seem to be OK.
But, could this (creating filesystems without partitioning the disks first) cause any performance or functional issues ?

I haven’t ‘farmed’ (equivalent term for mining in Chia world) any chia coins yet despite farming for 1 month. It may or may not be related how my Hard drives are formatted in the weird manner as mentioned above.

But, I wish Linux errored out when I ran $ sudo mkfs.ext4 /dev/sdc and forced me to partition the disks first. :pensive:

3 Likes
2

It won’t cause an error because it is a valid thing to do. It is how many people format removable media like flash drives. It is also common for virtualized or otherwise not directly on the hardware block devices.

That being said, I am not sure this has a performance penalty as much as a flexibility penalty. Have you tried testing the performance of the disks to see how they do?(While not farming)

Also, ext4 can be tuned, should ensure that you have the appropriate settings for your use case.

6 Likes
3

Thank You for the info, Dalto.

The PC should be farming 24/7, otherwise I lose opportunities to ‘sign the challenge’ and win coins.
But, now my PC and disks are not ‘plotting’ ; a CPU, memory, IO intensive activity where the hard drives are filled with ‘plot’ files. Plotting process is completed as I ran out of Hard drives and don’t have money to buy new hard drives :expressionless:

So, now I am only ‘farming’ which is less CPU/memory/disk IO intensive. So, I can still do some disk IO testing. Any popular way to test disk IO performance in Linux (fedora/Ubuntu) ?

4

Hey Rudolph

How about using dd command to do some basic disk IO test ?

https://www.cyberciti.biz/faq/howto-linux-unix-test-disk-performance-with-dd-command/