--- title: "Resize a Root ext4 Partition Online Without Reboot" source: "https://support.tools/post/online-resize-of-a-root-ext4-file-system/" published: 2022-07-14 categories: ["Ubuntu","Linux"] tags: ["Ubuntu","Linux","ext4","resize"] --- ## Quick fix After enlarging the virtual disk, rescan it, grow the partition and grow the file system while it stays mounted: `echo 1 > /sys/class/block/sda/device/rescan`, then `growpart /dev/sda 3` (or `parted /dev/sda resizepart 3 100%`), then `resize2fs /dev/sda3`. ext4 can only grow online; shrinking requires unmounting. The following is a guide to resizing your root ext4 file system online. After the operation, your partition will have more free space. There will be no shrinkage of the root file system. The root partition could have been any partition, not exactly the root one, but in most cases, such operations on the root are more complex and dangerous, so ALWAYS back up (or snapshot the VM) before performing such operations! During the resize operation, none of the services need to be shut down, rebooted, or unmounted. Even so, we rebooted the server once as a precaution, since it was possible and the server was not in production. Rebooting the server after this kind of resizing is not necessary. The method was tested on Ubuntu 22.04 with kernel 5.15 on a VMware virtual disk. Any distribution with a recent kernel, parted (or growpart) and e2fsprogs works the same way. Summary: - Make the kernel see the larger disk (rescan the SCSI device). - Grow the partition with `growpart` (package `cloud-guest-utils` on Debian and Ubuntu, `cloud-utils-growpart` on RHEL) or with `resizepart` in `parted`. - Grow the file system using `resize2fs` from [E2fsprogs](https://git.kernel.org/pub/scm/fs/ext2/e2fsprogs.git). There is a version of this package included in most Linux distributions with the same name. # [Steps](#steps) ## Detect disk size change Before we can resize the root file system, we need the OS to detect the new size of the disk. In this case, we use the `fdisk` command to show the current disk size. Which in this case is 120GB. (I know a 120GB root file system is not a good idea, but this is a lab.) ```bash root@a1ublabl01:~# fdisk -l /dev/sda Disk /dev/sda: 120 GiB, 128849018880 bytes, 251658240 sectors Disk model: Virtual disk Units: sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disklabel type: gpt Disk identifier: DDC91865-B329-443D-AFFF-0898C341D65C Device Start End Sectors Size Type /dev/sda1 2048 4095 2048 1M BIOS boot /dev/sda2 4096 4198399 4194304 2G Linux filesystem /dev/sda3 4198400 251656191 247457792 118G Linux filesystem root@a1ublabl01:~# ``` We now need to tell the OS to scan the disk for changes. For a single SCSI or virtual disk, write to its `rescan` file: ```bash echo 1 > /sys/class/block/sda/device/rescan ``` To rescan every disk, I use a script I created called [DiskScan](https://github.com/mattmattox/DiskScan). It walks through all SCSI hosts and disks and looks for changes. Read the script before piping it to `bash`. One-liner: ```bash curl https://raw.githubusercontent.com/mattmattox/DiskScan/master/rescan_disks.sh | bash ``` Output: ```bash root@a1ublabl01:~# curl https://raw.githubusercontent.com/mattmattox/DiskScan/master/rescan_disks.sh | bash % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 363 100 363 0 0 1553 0 --:--:-- --:--:-- --:--:-- 1551 Scan for new disk(s)... host0 host1 ... Scan for disk size change... sda root@a1ublabl01:~# ``` We can now see that the disk size has changed. Note, we can ignore the GPT error as we'll be fixing it as part of the next step. ```bash root@a1ublabl01:~# fdisk -l /dev/sda GPT PMBR size mismatch (251658239 != 419430399) will be corrected by write. The backup GPT table is not on the end of the device. Disk /dev/sda: 200 GiB, 214748364800 bytes, 419430400 sectors Disk model: Virtual disk Units: sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disklabel type: gpt Disk identifier: DDC91865-B329-443D-AFFF-0898C341D65C Device Start End Sectors Size Type /dev/sda1 2048 4095 2048 1M BIOS boot /dev/sda2 4096 4198399 4194304 2G Linux filesystem /dev/sda3 4198400 251656191 247457792 118G Linux filesystem root@a1ublabl01:~# ``` ## Resize the root partition At this point, the disk has the available space we need. We can now resize the root partition. ```bash root@a1ublabl01:~# parted /dev/sda GNU Parted 3.4 Using /dev/sda Welcome to GNU Parted! Type 'help' to view a list of commands. (parted) p Warning: Not all of the space available to /dev/sda appears to be used, you can fix the GPT to use all of the space (an extra 167772160 blocks) or continue with the current setting? Fix/Ignore? Fix Model: VMware Virtual disk (scsi) Disk /dev/sda: 215GB Sector size (logical/physical): 512B/512B Partition Table: gpt Disk Flags: Number Start End Size File system Name Flags 1 1049kB 2097kB 1049kB bios_grub 2 2097kB 2150MB 2147MB ext4 3 2150MB 129GB 127GB (parted) resizepart 3 -1 (parted) p Model: VMware Virtual disk (scsi) Disk /dev/sda: 215GB Sector size (logical/physical): 512B/512B Partition Table: gpt Disk Flags: Number Start End Size File system Name Flags 1 1049kB 2097kB 1049kB bios_grub 2 2097kB 2150MB 2147MB ext4 3 2150MB 215GB 213GB (parted) q Information: You may need to update /etc/fstab. root@a1ublabl01:~# ``` Note: `-1` is the new end of the partition, not a size. A negative value counts back from the end of the disk, so `-1` tells parted to use the remaining space. `100%` does the same thing. If `cloud-guest-utils` is installed, `growpart` does the same in one non-interactive command and also fixes the backup GPT header: ```bash growpart /dev/sda 3 ``` Note the space: `growpart` takes the disk and the partition number as two arguments. ## Resize the file system At this point, we have expanded the root partition to the new size. We can now resize the file system. ```bash root@a1ublabl01:~# resize2fs /dev/sda3 resize2fs 1.46.5 (30-Dec-2021) Filesystem at /dev/sda3 is mounted on /; on-line resizing required old_desc_blocks = 2, new_desc_blocks = 13 The filesystem on /dev/sda3 is now 51428620 (4k) blocks long. ``` Confirm the new size with `df -h /`. ## Optional: force a filesystem check on next reboot A check is not required after an online grow. If you want one anyway, note that `touch /forcefsck` is a SysV init convention that systemd-based releases such as Ubuntu 22.04 do not document. On systemd, add `fsck.mode=force` to the kernel command line for one boot (for example by editing the entry in the GRUB menu), or set a check on next mount with `tune2fs -c 1 -C 2 /dev/sda3` (maximum mount count 1, current count 2), then run `tune2fs -c -1 /dev/sda3` after the check to turn count-based checks off again. ## FAQ ### Can I shrink a mounted ext4 file system? No. `resize2fs` can grow a mounted ext4 file system, but shrinking requires the file system to be unmounted, which for `/` means booting from rescue media. ### My root file system is on LVM. What changes? Grow the partition that holds the physical volume as above, then run `pvresize /dev/sda3` and `lvextend -r -l +100%FREE /dev/ubuntu-vg/ubuntu-lv`. The `-r` flag runs `resize2fs` for you. Check your volume group and logical volume names with `lvs`. ### resize2fs says "The filesystem is already N blocks long. Nothing to do!" The partition was not grown, or the kernel still sees the old partition size. Check `lsblk /dev/sda` and, if the partition is still small, run `partprobe /dev/sda` or `partx -u /dev/sda` after resizing it. ### What about XFS? XFS also grows online, but with `xfs_growfs /` (it takes the mount point) instead of `resize2fs`. XFS cannot be shrunk at all. ## Related Guides - [Complete Linux LVM Management Guide](/linux-lvm-advanced-management-guide/) - [Linux Filesystem and Partition Management Guide](/linux-filesystem-partition-management-guide/) - [How to Copy GPT Partition Table Between Disks](/how-to-copy-gpt-partition-table/) - [Understanding XFS Inodes](/understanding-xfs-inodes/)