Quotas¶

Disk quotas cap the blocks and inodes a user, group or project may use on one filesystem. They stop one account or tenant from filling a shared volume.

Track: RHCSA · Interview weight: Low

Must-Know Facts¶

Fact	Value	Verify with
Limits	Soft (may be exceeded for the grace period, 7 days by default) and hard (never exceeded)	`quota -s <user>`
XFS	Mount options `uquota`, `gquota`, `pquota` at mount time; managed with `xfs_quota -x`	`findmnt -no OPTIONS <dir>`
ext4	`tune2fs -O quota -Q usrquota,grpquota` (unmounted), mount with `usrquota,grpquota`	`quotaon -p <dir>`
Set limits	`xfs_quota -x -c 'limit bsoft=50m bhard=60m <user>'`; `setquota -u <user> <bsoft> <bhard> <isoft> <ihard> <dir>`; `edquota -u <user>` opens an editor	`quota -s <user>`
Reports and grace	`xfs_quota -x -c 'report -h'`, `repquota -s <dir>`; grace with `xfs_quota -x -c 'timer ...'` or `setquota -t`	as root

XFS Quotas¶

XFS reads quota options only when the filesystem is mounted, so a remount ignores them without an error:

sudo mount -o remount,uquota /srv/web; echo "rc=$?"
findmnt -no OPTIONS /srv/web
sudo umount /srv/web
sudo mount -o uquota,gquota /dev/vgdata/lvweb /srv/web
findmnt -no OPTIONS /srv/web
sudo mkdir /srv/web/uploads && sudo chown laborant: /srv/web/uploads
sudo xfs_quota -x -c 'limit bsoft=50m bhard=60m isoft=900 ihard=1000 laborant' /srv/web
dd if=/dev/zero of=/srv/web/uploads/a.bin bs=1M count=80 status=none; echo "rc=$?"
sudo xfs_quota -x -c 'report -h -u' /srv/web

Output:

rc=0
rw,relatime,attr2,inode64,logbufs=8,logbsize=32k,noquota
rw,relatime,attr2,inode64,logbufs=8,logbsize=32k,usrquota,grpquota
dd: error writing '/srv/web/uploads/a.bin': Disk quota exceeded
rc=1
User quota on /srv/web (/dev/mapper/vgdata-lvweb)
                        Blocks              
User ID      Used   Soft   Hard Warn/Grace   
---------- --------------------------------- 
root            0      0      0  00 [------]
laborant      60M    50M    60M  00 [------]

The write stopped at the 60 MiB hard limit. The permanent form is defaults,uquota,gquota in the fstab line; for the root filesystem, rootflags=uquota goes on the kernel command line.

Quota options on an XFS remount are ignored

mount -o remount,uquota returns 0 and leaves noquota in place. Unmount and mount again, or reboot after editing fstab.

ext4 Quotas¶

sudo umount /srv/db
sudo tune2fs -O quota -Q usrquota,grpquota /dev/vgdata/lvdb | tail -1
sudo mount -o usrquota,grpquota /dev/vgdata/lvdb /srv/db
sudo mkdir -p /srv/db/home && sudo chown laborant: /srv/db/home
sudo setquota -u laborant 20M 25M 0 0 /srv/db
dd if=/dev/zero of=/srv/db/home/b.bin bs=1M count=22 status=none; echo "rc=$?"
dd if=/dev/zero of=/srv/db/home/c.bin bs=1M count=10 status=none; echo "rc=$?"
sudo repquota -s /srv/db

Output:

tune2fs 1.47.1 (20-May-2024)
rc=0
dd: error writing '/srv/db/home/c.bin': Disk quota exceeded
rc=1
*** Report for user quotas on device /dev/mapper/vgdata-lvdb
Block grace time: 7days; Inode grace time: 7days
                        Space limits                File limits
User            used    soft    hard  grace    used  soft  hard  grace
----------------------------------------------------------------------
root      --     24K      0K      0K              3     0     0       
laborant  +-  25600K  20480K  25600K  7days       3     0     0

In repquota, +- means over the block soft limit and within the inode limits. The quota feature keeps usage in hidden inodes, so the old quotacheck and aquota.user files are not needed.

Passing the soft limit starts the grace period

The first file passed the soft limit and succeeded, which started the 7-day grace period; the second hit the hard limit. After the grace time, the soft limit is enforced like a hard one.

Interview Checkpoints¶

L1: What is the difference between a soft and a hard quota?

Say first: the hard limit is never exceeded; the soft limit may be exceeded for the grace period, after which it is enforced like a hard limit.

Proof: quota -s <user> shows usage, both limits and the grace time.

Follow-up: What error does a program get at the limit? (EDQUOT, Disk quota exceeded.)

L2: Limit user dev to 500 MiB on the XFS volume /data.

Say first: mount with uquota, then set the limit with xfs_quota.

Proof: fstab defaults,uquota; sudo umount /data; sudo mount /data; sudo xfs_quota -x -c 'limit bsoft=450m bhard=500m dev' /data.

Follow-up: Why did mount -o remount,uquota not work?

L2: Show every user's usage and limits on an ext4 filesystem.

Say first: repquota.

Proof: sudo repquota -s /home.

Follow-up: How do you copy one user's limits to others? (edquota -p <proto> <user>....)

L2: Limit a directory tree rather than a user on XFS.

Say first: use a project quota: mount with pquota, assign the directory a project ID and set the limit on the project.

Proof: sudo xfs_quota -x -c 'project -s -p /data/app 42' -c 'limit -p bhard=1g 42' /data.

Follow-up: Which container runtimes use XFS project quotas? (Docker and CRI-O for per-container storage limits on overlay2 over XFS.)

L3: A user gets Disk quota exceeded, but df shows plenty of space. What do you check?

Say first: the user's quota on that filesystem, including files owned in other directories and the grace time.

Proof: quota -s <user>; sudo find <mount> -xdev -user <user> -size +10M.

Follow-up: Which group quota could also cause it?

L2: Enable quotas on an existing ext4 volume.

Say first: unmount, turn on the quota feature with tune2fs, and mount with the quota options.

Proof: sudo tune2fs -O quota -Q usrquota /dev/<dev>; usrquota in fstab; sudo quotaon -p <mount>.

Follow-up: What did older setups need instead? (quotacheck -cug and quotaon.)

Disk Usage: full filesystems and inodes; Mounting and fstab: mount options

Captured on Rocky Linux 10.2 (iximiuz Labs microVM, kernel 6.1.167, quota 4.09), 2026-09.