AWS EC2 /dev/nvme1n1 mounted without partition table

Question

I have an EC2 server on nitro system (c6g.2xl) where I have attached a 2nd volume. This is a web hosting environment and the disk is regularly increased in size.

For some reason I accidentally mounted the disk on - what I would call - the "partition table level", ie. my mount point is /dev/nvme1n1, not /dev/nvme1n1p1 - I can't recall how this happened, but it has the following effects:

when I make the disk bigger, it is immediately bigger - I do not need to grow the partition table. In fact when I try growpart as per https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/recognize-expanded-volume-linux.html I get an error:
```
sfdisk: /dev/nvme1n1: does not contain a recognized partition table
FAILED: failed to dump sfdisk info for /dev/nvme1n1
```
I have successfully restored from a snapshot by attaching it as a 3rd disk

Given that seemingly all of the Internet has a partition table on this volume - what are the downsides of this approach?

At my place of employment we don't use partition tables for data volumes. Some far-fetched downsides I can think of are (a) "smart" programs that manipulate disks thinking the volume is uninitialized and helpfully automatically writing a partition table to the disk and (b) novice sysadmins doing the same. — Mark Wagner, Aug 09 '22 at 00:05

score 2 · Answer 1 · answered Aug 08 '22 at 22:06

2

There is no need for a partition table on a data disk, it's only necessary if you want to boot from it. If you want to split a volume into smaller block devices, I'd suggest using LVM.

answered Aug 08 '22 at 22:06

fuero

9,413
1
35
40

and would you view most guides as using a partition table though? Do you have an opinion why they do? – jdog Aug 08 '22 at 22:59
2

Mostly inexperience and a little bit of cargo-culting I guess. – fuero Aug 09 '22 at 11:14

AWS EC2 /dev/nvme1n1 mounted without partition table

1 Answers1