Partitioning a Rocky Linux 9 EC2

Rocky Linux Tech Help
1

I have a strange issue while working with partitions on my Rocky Linux 9 (non LVM) EC2 instance on AWS.

GOAL: I would like to enlarge a little (5GB) my ebs in order to create a new partition and a new xfs filesystem to mount on it /tmp to have it separated from root partition

MY ATTEMPT:
I started to remove from /etc/cloud/cloud.cfg growpart and resizefs to prevent cloud-init to automatically enlarge root partition:

sudo sed -i ‘/ - growpart/d’ /etc/cloud/cloud.cfg
sudo sed -i ‘/ - resizefs/d’ /etc/cloud/cloud.cfg

Then I enlarged my EBS volume from AWS console, from 15GB to 20GB

After that on my ec2 I see, from lsblk ,that the disk has the desired size, so I started my work

  • I created a new partition with fdisk
  • I created a filesystem on it: mkfs.xfs -f -L tmp /dev/nvme0n1p5

Until here, every reboot works

  • I modified /etc/fstab to mount at boot the partition
    LABEL=tmp /tmp xfs defaults,noatime,nodev,nosuid,noexec 1 2
  • then I re-mount (mount -a) to ensure everything works

Tmp is now on my new paritition

Then I reboot my ec2 and the OS is stuck, I cannot connect to it anymore

QUESTIONS:
What is wrong with all of this? What am I missing? tmpfs?
I just need a separated partition, only one this time but can be more in the future, is there a guide to to this on aws?

Thanks

2

Maybe try removing the fsck flag. fsck cannot run on xfs. Unless your system has been configured for this flag to trigger xfs_repair, it should just execute fsck.xfs, which does nothing:

System administrator can force fsck.xfs to run xfs_repair(8) at boot time by creating a /forcefsck file or booting the system with “fsck.mode=force” on the kernel command line.
So, fsck.xfs will do nothing at all.
Default /etc/fstab settings for / partition - Red Hat Customer Portal

But maybe for some reason it causes the error?

Anyway, is there a reason why you set the dump flag to 1? As far as I know, dump is rarely used these days and incompatible with xfs. To my understanding setting it should have no effect on xfs.

3

I think I found the issue and the root cause is tmp.mount. In many guide I found online that to mount tmp with an entry in fstab I need to disable that service with
systemctl mask tmp.mount
With this command, that should disable the service, the corresponding unit is redirect to null, via symbolic link I suppose

[root@rocky9 ~]# ls -l /etc/systemd/system/tmp.mount
lrwxrwxrwx. 1 root root 9 May 23 17:20 /etc/systemd/system/tmp.mount → /dev/null
[root@rocky9 ~]# cat /etc/systemd/system/tmp.mount
[root@rocky9 ~]#

Operation that seemed useless since on Rocky 9 the configuration is that one by default.

Only in this page

I found the command
[root@rocky9 ~]# systemctl unmask tmp.mount
Removed “/etc/systemd/system/tmp.mount”.

That completely remove the file and I’m able to mount via fstab.

I don’t know exactly the meaning and the action of mask and unmask commands, but that’s an other topic.

1 Like
4

If you mask the service, it cannot be enabled/disabled or started/restarted/stopped until it has been unmasked.