repartitioning in-use, contiguous space in Linux. To recap the
situational setup, a filesystem (FS) has run out of space and the data
contained on it couldn't be removed or compressed to free up space.
Since the next adjacent partition is configured as swap space, the plan
is to remove that partition and grow our fully consumed FS into space
reclaimed from the original swap partition while leaving our configured
FS intact. The host details for our scenario are:
HOST: europa
PROMPTS: multi user: europa [0]
single user: europa-cons [0]
OS: CentOS 6.2 (RHEL variant)
FS TYPE: ext4
DISK: sdb
NOTE: I've also used the following with no issues on
previous versions of CentOS (and RHEL) with ext3
filesystems as well.
Before continuing, a similar resolution to this scenario could havebeen to simply extend the volume using LVM. Since our volumes on this
disk are not under LVM control, this is not an easily accessible option,
thus reclaiming the adjacent partition was chosen instead.
Starting things off, our application resides solely on disk "sdb" on
partitions "1" and "3", mounted under "/opt/myapp". There is also a swap
partition at "sdb2". Below, our 'df' output shows the FS in question on
"sdb1" (at 100% capacity), our swap volumes, and the partition table on
"sdb" as reported by 'fdisk':
europa [0] /bin/df -h
Filesystem Size Used Avail Use% Mounted on
/dev/sda1 6.0G 1.5G 4.2G 27% /
tmpfs 467M 0 467M 0% /dev/shm
/dev/sda3 1008M 97M 861M 11% /var
/dev/sdb1 510M 510M 0 100% /opt/myapp/logs
/dev/sdb3 510M 8.9M 475M 2% /opt/myapp/bin
europa [0] /sbin/swapon -s
Filename Type Size Used Priority
/dev/sda2 partition 1048568 0 -1
/dev/sdb2 partition 530136 0 -2
europa [0] /sbin/fdisk -l /dev/sdb
Disk /dev/sdb: 2147 MB, 2147483648 bytes
255 heads, 63 sectors/track, 261 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x9ae7b0fc
Device Boot Start End Blocks Id System
/dev/sdb1 1 66 530113+ 83 Linux
/dev/sdb2 67 132 530145 82 Linux swap / Solaris
/dev/sdb3 133 198 530145 83 Linux
europa [0]
Since "sdb2" is swap space, and in this case, we have additional space atthe end of the disk, we are going to repartition this disk. The result
will be that partitions "sdb1" and "sdb2" will be removed and "sdb1"
recreated (starting at the same cylinder) but this time, encompassing
the space previously held by "sdb2". Next, "sdb2" will be recreated,
this time using the available space at the end of the disk (due to the
size of this disk, "sdb2 (our swap volume) will be about 23 MB smaller
once reconfigured). To prepare for this, using your favorite editor
(vi?) update the entries for "sdb" in "/etc/fstab" from this:
europa [0] /bin/grep sdb /etc/fstab
/dev/sdb2 swap swap defaults 0 0
/dev/sdb1 /opt/myapp/logs ext4 defaults 1 2
/dev/sdb3 /opt/myapp/bin ext4 defaults 1 2
europa [0]
to this:
europa [0] /bin/grep sdb /etc/fstab
#/dev/sdb2 swap swap defaults 0 0
#/dev/sdb1 /opt/myapp/logs ext4 defaults 1 2
#/dev/sdb3 /opt/myapp/bin ext4 defaults 1 2
europa [0]
Once complete, we'll need to reboot into "single user mode". This isto ensure that the "sdb" FS are not in use and removes "sdb2" from swap
usage. (If you can otherwise stop I/O to the "sdb" FS and remove "sdb2"
from swap usage, you could technically do this in "multi user mode".):
europa [0] /sbin/reboot
Broadcast message from root@europa
(/dev/tty1) at 13:42 ...
The system is going down for reboot NOW!
europa [0]
(screen refresh)
Running guests on default URI: libvirtd not installed; skipping this URI.
Stopping atd: [ OK ]
<snip...>
Sending all processes the TERM signal... [ OK ]
Sending all processes the KILL signal... [ OK ]
Saving random seed: [ OK ]
Syncing hardware clock to system time [ OK ]
Turning off swap: [ OK ]
Turning off quotas: [ OK ]
Unmounting file systems: [ OK ]
init: Re-executing /sbin/init [ OK ]
Please stand by while rebooting the system...
md: stopping all md devices.
sd 3:0:0:0: [sdb] Synchronizing SCSI cache
sd 2:0:0:0: [sda] Synchronizing SCSI cache
Once the host has rebooted and after the BIOS POSTs, we need to editour GRUB boot entry. Select the appropriate entry (on this host, there
is only one) and hit 'e' before the 5 seconds time out so that we can
update the kernel boot parameters:
GNU GRUB version 0.97 (639K lower / 1047488K upper memory)
CentOS 6.2 (2.6.32-220.el6.x86_64
Use the <up> and <down> keys to select which entry is highlighted.
Press enter to boot the select OS, 'e' to edit the
commands before booting, or 'c' for a command-line.
The highlighted entry will be booted automatically in 5 seconds.
(after entry selection and hitting 'e'):
GNU GRUB version 0.97 (639K lower / 1047488K upper memory)
root (hd0,0)
kernel /boot/vmlinux-2.6.32-220.el6.x86_64 ro root=UUID=05e46d98-6e82->
initrd /boot/initramfs-2.6.32-220.el6.x86_64.img
Use the <up> and <down> keys to select which entry is highlighted.
Prebb 'b' to boot, 'e' to edit the selected command in the
boot sequence, 'c' for a command-line, 'o' to open a new line
after ('O' for before) the selected line, 'd' to remove the
selected line, or escape to go back to the main menu.
Navigate to the 'kernel' line, hit 'e', and update it from this:
kernel /boot/vmlinux-2.6.32-220.el6.x86_64 <...> rd_NO_DM
to this:
kernel /boot/vmlinux-2.6.32-220.el6.x86_64 <...> rd_NO_DM -s
(Since you may have other options set for your kernel, the intent hereis to update the entry to include "-s" at the end so that we boot into
"single user mode".) After hitting 'Enter', type 'b' to boot the system,
at which point the screen will refresh and we'll see the boot process:
<snip...>
Enabling /etc/fstab swaps: [ OK ]
root@europa /]# _
Log in at the prompt above. Once logged in, I've simply reset PS1 forclaritiy of this writeup:
root@europa /]# PS1=`/bin/hostname -s`'-cons [$?] '
europa-cons [0]
europa-cons [0] /usr/bin/who -r
run-level S 2013-01-17 13:49
europa-cons [0] /sbin/runlevel
N S
europa-cons [0] /bin/df -h
Filesystem Size Used Avail Use% Mounted on
/dev/sda1 6.0G 1.5G 4.2G 27% /
tmpfs 467M 0 467M 0% /dev/shm
/dev/sda3 1008M 97M 861M 11% /var
europa-cons [0] /sbin/swapon -s
Filename Type Size Used Priority
/dev/sda2 partition 1048568 0 -1
europa-cons [0]
Above, I've also checked our current run level via 'who -r' (and via'runlevel' for those who didn't learn on Solaris). Additionally, I've
run 'df' and 'swapon' to simply verify that "sdb" is no longer in use.
At this point, we get to repartition "sdb". It must be stressed before
we do any repartitioning, either we must evacuate the adjacent partition
of data (if it contains an FS), the adjacent partition was in use but
can be reclaimed, or subsequent contiguous space must exist to our
partition that we're concerned with (sdb1). If this is not the case,
you will likely risk data loss if you proceed any further.
Moving along, I've called 'fdisk /dev/sdb' to operate on that disk and
verified the existing partition table via "p":
europa-cons [0] /sbin/fdisk /dev/sdb
WARNING: DOS-compatible mode is deprecated. It's strongly recommended to
switch off the mode (command 'c') and change display units to
sectors (command 'u').
Command (m for help): p
Disk /dev/sdb: 2147 MB, 2147483648 bytes
255 heads, 63 sectors/track, 261 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x9ae7b0fc
Device Boot Start End Blocks Id System
/dev/sdb1 1 66 530113+ 83 Linux
/dev/sdb2 67 132 530145 82 Linux swap / Solaris
/dev/sdb3 133 198 530145 83 Linux
Our first order of business is to zero out partition 2 (sdb2, theswap partition). After that, I've also removed partition 1 (sdb1)
and subsequently recreated it setting the first cylinder to be "1"
(as it was originally). I've also set partition 1 to end on cylinder
"132", just as partition 2 originally did. Finally, I've re-added
our swap space back to partition 2, setting the starting cylinder to
be next available (199) and accepting maximum value available (261),
and set the partition type to 82 (Linux swap / Solaris):
Command (m for help): d
Partition number (1-4): 2
Command (m for help): d
Partition number (1-4): 1
Command (m for help): n
Command action
e extended
p primary partition (1-4)
p
Partition number (1-4): 1
First cylinder (1-261, default 1): 1
Last cylinder, +cylinders or +size{K,M,G} (1-132, default 132): 132
Command (m for help): n
Command action
e extended
p primary partition (1-4)
p
Partition number (1-4): 2
First cylinder (199-261, default 199):
Using default value 199
Last cylinder, +cylinders or +size{K,M,G} (199-261, default 261): +512M
Value out of range.
Last cylinder, +cylinders or +size{K,M,G} (199-261, default 261):
Using default value 261
Command (m for help): t
Partition number (1-4): 2
Hex code (type L to list codes): 82
Changed system type of partition 2 to 82 (Linux swap / Solaris)
Of note, I initially tried setting the partion size for partition 2 to"512M". Since the disk didn't have that much space available, I instead
settled for about 23 MB less and accepted the default end cylinder (261).
Below, I've confirmed our new partition table (p) and labeled the disk
(wrote the partition table to the disk):
Command (m for help): p
Disk /dev/sdb: 2147 MB, 2147483648 bytes
255 heads, 63 sectors/track, 261 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x9ae7b0fc
Device Boot Start End Blocks Id System
/dev/sdb1 1 132 1060258+ 83 Linux
/dev/sdb2 199 261 506047+ 82 Linux swap / Solaris
/dev/sdb3 133 198 530145 83 Linux
Partition table entries are not in disk order
Command (m for help): w
The partition table has been altered!
Calling ioctl() to re-read partition table.
Syncing disks.
We can now uncomment our "sdb" related "fstab" entries, so using yourfavorite text editor, update "fstab" from this:
europa [0] /bin/grep sdb /etc/fstab
#/dev/sdb2 swap swap defaults 0 0
#/dev/sdb1 /opt/myapp/logs ext4 defaults 1 2
#/dev/sdb3 /opt/myapp/bin ext4 defaults 1 2
to this:
europa [0] /bin/grep sdb /etc/fstab
/dev/sdb2 swap swap defaults 0 0
/dev/sdb1 /opt/myapp/logs ext4 defaults 1 2
/dev/sdb3 /opt/myapp/bin ext4 defaults 1 2
For Linux to use the newly recreated swap space, below, I've run 'mkswap'against "/dev/sdb2". Also, while their shouldn't be any issue, I've
verified the sanity of our contained FS on "sdb1" using 'fsck':
europa-cons [0] /sbin/mkswap /dev/sdb2
Setting up swapspace version 1, size = 506040 KiB
no label, UUID=c93960d1-aaa3-49de-98b5-eca5803c27ea
europa-cons [0] /sbin/fsck -y /dev/sdb1
fsck from util-linux-ng 2.17.2
e2fsck 1.41.12 (17-May-2010)
/dev/sdb1: clean, 16729/33200 files, 132528/132528 blocks
If our FS checks out to be clean (which it does), below we 'mount'the "sdb1" FS to "/opt/myapp/logs", run 'df' to see the existing size
and usage, extend the EXT4 FS on "sdb1" using 'resize2fs', and finally
recheck our size and usage via 'df':
europa-cons [0] /bin/mount /opt/myapp/logs
europa-cons [0] /bin/df -h /opt/myapp/logs
Filesystem Size Used Avail Use% Mounted on
/dev/sdb1 510M 510M 0 100% /opt/myapp/logs
europa-cons [0] /sbin/resize2fs -p /dev/sdb1
resize2fs 1.41.12 (17-May-2010)
Filesystem at /dev/sdb1 is mounted on /opt/myapp/logs; on-line resizing required
old desc_blocks = 1, new_desc_blocks = 1
Performing an on-line resize of /dev/sdb1 to 265064 (4k) blocks.
The filesystem on /dev/sdb1 is now 265064 blocks long.
europa-cons [0] /bin/df -h /opt/myapp/logs
Filesystem Size Used Avail Use% Mounted on
/dev/sdb1 1021M 510M 466M 53% /opt/myapp/logs
europa-cons [0] /bin/ls /opt/myapp/logs
account crash db games local log mail opt preserve spool yp
cache cvs empty lib lock lost+found nis other run tmp
(The "-p" option to 'resize2fs' will print out percentage of completionwhile 'resize2fs' is working, which given the size of our FS, is
irrelevant.) Above, we see in the second 'df' that our FS size has indeed
been extended to consume the space reclaimed from "sdb1". As a follow up,
our files are still accessible as seen in the 'ls'. Below, we 'umount'
"/opt/myapp/logs" and run 'fsck' once more on "sdb1" to verify the FS
sanity as a follow up to our extending the FS with 'resize2fs' above:
europa-cons [0] /bin/umount /opt/myapp/logs
europa-cons [0] /sbin/fsck -y /dev/sdb1
fsck from util-linux-ng 2.17.2
e2fsck 1.41.12 (17-May-2010)
/dev/sdb1: clean, 16729/59760 files, 134264/265064 blocks
europa-cons [0] /sbin/reboot
(screen refresh)
Sending all processes the TERM signal... [ OK ]
Sending all processes the KILL signal... init: rsS-sulogin main process (701)\
killed by KILL signal
[ OK ]
Saving random seed: [ OK ]
Syncing hardware clock to system time [ OK ]
Turning off swap: [ OK ]
Turning off quotas: [ OK ]
Unmounting file systems: [ OK ]
init: Re-executing /sbin/init [ OK ]
Please stand by while rebooting the system...
md: stopping all md devices.
sd 3:0:0:0: [sdb] Synchronizing SCSI cache
sd 2:0:0:0: [sda] Synchronizing SCSI cache
Providing our FS is clean, above we reboot the host. This time, we don'tneed to modify the GRUB boot entry below, so just let the menu time out
and boot up the host to the default run level (in this case, 3):
GNU GRUB version 0.97 (639K lower / 1047488K upper memory)
CentOS 6.2 (2.6.32-220.el6.x86_64
Use the <up> and <down> keys to select which entry is highlighted.
Press enter to boot the select OS, 'e' to edit the
commands before booting, or 'c' for a command-line.
The highlighted entry will be booted automatically in 5 seconds.
(screen refresh)
CentOS release 6.2 (Final)
Kernel 2.6.32-220.el6.x86_64 on an x86_64
europa login: _
With our host now back online, we simply verify that "sdb2" is allocatedto swap using 'swapon -s' and our "sdb" FS are mounted:
europa [0] /sbin/swapon -s
Filename Type Size Used Priority
/dev/sda2 partition 1048568 0 -1
/dev/sdb2 partition 506036 0 -2
europa [0] /bin/df -h
Filesystem Size Used Avail Use% Mounted on
/dev/sda1 6.0G 1.5G 4.2G 27% /
tmpfs 467M 0 467M 0% /dev/shm
/dev/sda3 1008M 97M 861M 11% /var
/dev/sdb1 1021M 510M 466M 53% /opt/myapp/logs
/dev/sdb3 510M 8.9M 475M 2% /opt/myapp/bin
europa [0]
Our work is now complete. For the curious, the reason we can re-layoutthe partition table in this manner is that a partition can be treated
as simply a container for the contained FS. Providing that you are not
reducing the size of the partition (container) or shifting the start/end
of the slice boundaries into space occupied by an FS, your FS is not
otherwise impacted and remains sane.
see also:
Repartitioning Contiguous Space in Solaris
