When doing a standard Linux installation on a XenServer virtual machine, the installation process typically allocates all of the available virtual disk space to the Linux file system and swap partitions. The size of the virtual disk can easily be changed using the XenCenter utility, but as neither the partition table or the file systems themselves are modified by XenCenter, the guest system does not see the new virtual disk size.

This article describes how to safely change the partitions.

There are two scenarios:

  • the Linux system uses standard Linux partitions
  • the Linux system uses Logical Volume Manager (LVM) partitions

The file system sits on top of a single block of data storage, such as a piece of the harddisk or, in the case of a XenServer virtual machine, a virtual block device. It can only be as large as this storage block, but not larger. Therefore, if the file system is to be increased, this storage block must be increased first.

After the size of the storage block has been increased, execute the resize2fs command, which can be safely run even on a mounted filesystem.

It is very simple to increase the size of the underlying storage block if you are using LVM (and this is a good reason to always use LVM). Just add more blocks to it using the lvresize command.

The blocks added to the storage block are taken from the current volume group. If the volume group does not have free block, new blocks must be made available by adding a new physical device such as a second harddisk, or by increasing an existing physical device.

Provide more disk space to the guest system

There are two ways to do this: you could either increase the size of the existing XenServer virtual disk, or add a second virtual disk to this guest.

Add a physical device

(todo)

Increase the size of an existing XenServer virtual device

The XenCenter administration software provides a dialog to change the size of the XenServer virtual disk. Select the new storage size by entering the value in the dialog. The virtual disk will be resized to the new value. When increasing the size, the data on the disk is preserved.

Note: do not decrease the size without decreasing the file system first. Otherwise the file system will be damaged.

Increasing the size of the XenServer virtual disk size does not change the filesystem on that disk. So the new disk space is not yet visible to the user. To use the new disk space, the partition table must be modified (which makes the new device size known to the file system) and the file system must be extended (to use the extra space).

The easiest way to do this is delete the last partition and create it again which will then use all of the remaining disk space by default. Obviously, this will work only if the last partition is in fact the partition that is to be increased.

In the example below, the XenServer virtual disk size has been resized from 10 GByte to 50 GByte. The following steps will be performed with fdisk:

  • print old partitions
  • delete last partition (#2)
  • create new partition using default settings (use the whole virtual disk for this partition)
  • set partition type to LVM
  • print partitions settings again, note increased size for partition #2
  • reboot
master:~# fdisk /dev/xvda

The number of cylinders for this disk is set to 6527.
There is nothing wrong with that, but this is larger than 1024,
and could in certain setups cause problems with:
1) software that runs at boot time (e.g., old versions of LILO)
2) booting and partitioning software from other OSs
   (e.g., DOS FDISK, OS/2 FDISK)

Command (m for help): p

Disk /dev/xvda: 53.6 GB, 53687091200 bytes     <-- note that fdisk already sees the new device size of 53.6 GB
255 heads, 63 sectors/track, 6527 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x0004e475

    Device Boot      Start         End      Blocks   Id  System
/dev/xvda1   *           1          31      248976   83  Linux
/dev/xvda2              32        1305    10233405   8e  Linux LVM     <-- note that the existing entry is still 10 GB

Command (m for help): d
Partition number (1-4): 2

Command (m for help): n
Command action
   e   extended
   p   primary partition (1-4)
p
Partition number (1-4): 2
First cylinder (32-6527, default 32):
Using default value 32
Last cylinder or +size or +sizeM or +sizeK (32-6527, default 6527):
Using default value 6527

Command (m for help): t
Partition number (1-4): 2
Hex code (type L to list codes): 8e
Changed system type of partition 2 to 8e (Linux LVM)

Command (m for help): p

Disk /dev/xvda: 53.6 GB, 53687091200 bytes
255 heads, 63 sectors/track, 6527 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x0004e475

    Device Boot      Start         End      Blocks   Id  System
/dev/xvda1   *           1          31      248976   83  Linux
/dev/xvda2              32        6527    52179120   8e  Linux LVM     <-- note that the new entry is now 50 GB

Command (m for help): w
The partition table has been altered!

Calling ioctl() to re-read partition table.

WARNING: Re-reading the partition table failed with error 16: Device or resource busy.
The kernel still uses the old table.
The new table will be used at the next reboot.
Syncing disks.
master:~# reboot

Broadcast message from root@master (pts/0) (Fri Oct  2 08:47:05 2009):

The system is going down for reboot NOW!

Tell LVM of the new resized physical volume

The partition #2 has now successfully increased in size, now let LVM use the new blocks. Note how the "Free PE" value increases after the pvresize command.

 pvdisplay
  --- Physical volume ---
  PV Name               /dev/xvda2
  VG Name               dsk
  PV Size               9.76 GB / not usable 1.37 MB
  Allocatable           yes (but full)
  PE Size (KByte)       4096
  Total PE              2498
  Free PE               0
  Allocated PE          2498
  PV UUID               7TyUyk-hyVf-0kdC-OCok-QdSJ-lqTE-YFJAi0

master:~# pvresize /dev/xvda2
  Physical volume "/dev/xvda2" changed
  1 physical volume(s) resized / 0 physical volume(s) not resized

master:~# pvdisplay
  --- Physical volume ---
  PV Name               /dev/xvda2
  VG Name               dsk
  PV Size               49.76 GB / not usable 3.98 MB
  Allocatable           yes
  PE Size (KByte)       4096
  Total PE              12738
  Free PE               10240
  Allocated PE          2498
  PV UUID               7TyUyk-hyVf-0kdC-OCok-QdSJ-lqTE-YFJAi0

Resize logical volume

The next step is to add those new blocks to the logical volume. The volume group now has an extra 10240 free extents (each 4 KByte in size). New storage can be added to a logical volume with the lvresize command.

# vgdisplay
  --- Volume group ---
  VG Name               dsk
  System ID
  Format                lvm2
  Metadata Areas        1
  Metadata Sequence No  8
  VG Access             read/write
  VG Status             resizable
  MAX LV                0
  Cur LV                2
  Open LV               2
  Max PV                0
  Cur PV                1
  Act PV                1
  VG Size               49.76 GB
  PE Size               4.00 MB
  Total PE              12738
  Alloc PE / Size       2498 / 9.76 GB
  Free  PE / Size       10240 / 40.00 GB
  VG UUID               nCjL0f-XWps-G77Z-Oq0t-lMEc-ULsJ-2qFUnw

master:~# lvdisplay
  --- Logical volume ---
  LV Name                /dev/dsk/root
  VG Name                dsk
  LV UUID                QejNO3-8bB0-ykJ0-TUhR-56Ko-W5ed-SvF9fv
  LV Write Access        read/write
  LV Status              available
  # open                 1
  LV Size                7.76 GB
  Current LE             1986
  Segments               1
  Allocation             inherit
  Read ahead sectors     auto
  - currently set to     256
  Block device           254:0

master:~# lvresize -l +10240 /dev/dsk/root
  Extending logical volume root to 47.76 GB
  Logical volume root successfully resized

master:~# vgdisplay
  --- Volume group ---
  VG Name               dsk
  System ID
  Format                lvm2
  Metadata Areas        1
  Metadata Sequence No  11
  VG Access             read/write
  VG Status             resizable
  MAX LV                0
  Cur LV                2
  Open LV               1
  Max PV                0
  Cur PV                1
  Act PV                1
  VG Size               49.76 GB
  PE Size               4.00 MB
  Total PE              12738
  Alloc PE / Size       12738 / 49.76 GB
  Free  PE / Size       0 / 0
  VG UUID               nCjL0f-XWps-G77Z-Oq0t-lMEc-ULsJ-2qFUnw

Resize file system

The last step is to resize the file system. This can be done even when the file system is currently in use. However, a reboot is recommended as massive changes have been made to the file system. During reboot, a file system check is executed which will print out a scary warning message but that's normal. The system will reboot again and this time no messages or warnings should be sent.

master:~# resize2fs /dev/dsk/root
resize2fs 1.41.3 (12-Oct-2008)
Filesystem at /dev/dsk/root is mounted on /; on-line resizing required
old desc_blocks = 1, new_desc_blocks = 3
Performing an on-line resize of /dev/dsk/root to 12519424 (4k) blocks.
The filesystem on /dev/dsk/root is now 12519424 blocks long.

master:~# reboot

Broadcast message from root@master (pts/0) (Fri Oct  2 08:47:05 2009):

The system is going down for reboot NOW!

20 Comments

  1. Anonymous

    Thanks for this great article. Helped me a great deal.

    David.

  2. Anonymous

    Great article, very accurate and thorough! Thanks!

  3. Anonymous

    How do i add some partitions in xenserver host to store my template backup of the VM? 

    because by default, Xen only creates 4GB of its root which is insufficient in storing backup files?

    Hope you may help me....

    Thanks! 

    1. Administrator

      The backups should not be stored on the server anyway. In case of a disk or server crash the backups would be inaccessible.

      I am using an NFS server to backup VMs:

      mkdir /backup
mount mybackupserver.mydomain:/some/storage /backup
...

      You need of course a fast network connection to do this as the backup files tend to be very large.

  4. Anonymous

    that is cool!  thanks.

  5. Anonymous

    Thanks a lot!!! :-)

  6. Anonymous

    Most Excellent documenting of the steps needed.  This saved me a lot of head-scratching to figure out all the steps end-to-end in resizing a logical volume!

    Michael

  7. Anonymous

    Hi,

    Thanks for this excellent article.

    I'm wondering. Deleting the partition table with fdisk, won't this make all the data of the partition unreadable after the resize (like a new format) ?

    In my case, I've got a server with an LVM partition (210 GB). I created 2 disks (/dev/mapper/xend1; /dev/mapper/xend2) each 100 GB.

    I created 2 virtual machines assigning them xend1, xend2 as hard disks (installing the virtual os on those disks).

    First I extended the disk space of my LVM volume group, then, with lvextend I increased the size of xend1. But, as you explained in your article, my vm does not yet see the new disk space.

    My concern is when applying the steps explained in your article, that I will have to completely reinstall my vm.

    Cheers,

    Christian

    1. Administrator

      When deleting a partition with fdisk, only the partition table is deleted. The actual data is still in place, but it would not be accessible by the standard OS commands. When you add a new partition right after deleting the partition, the new partition will point to the same data blocks, but will have lots of extra blocks after the end of the existing file system (note that this only works if the deleted and recreated partition is the last partition).

      However, if you are using LVM anyway, the easiest solution would be to simply create a new virtual device, which will show up as /dev/xvdb, for example. This new virtual disk can then be used to add more blocks to a volume group, so that you can use lvesize as described. See also How to add additional disks to XenServer host.

  8. Anonymous

    Too many comlipmnets too little space, thanks!

  9. Anonymous

    Thanks - good stuff.

  10. Anonymous

    Thanks so much for taking the time to write this very helpful guide. All the best! you helped me a lot.

    Just wanted to help anyone out there who might get condused - When you're resizing the logical volume, make sure you do it to the right volume! I got confused and didn't do this on the root volume. (On my Xen machine this is located under /dev/mapper/vg_vmname-lv_root) - and ran into problems when trying to resize the file system.

    Thanks again! I'll bookmark this so I can return to it.

  11. Anonymous

    Thanks.. Its perfect steps..

     

    Warm Regards

    Rupesh Kumar

  12. Anonymous

    Much thanks for this article, it was step by step exactly what I needed!

  13. Anonymous

    Thank you!

    This article was clear, precise and correct! Bravo!

  14. Anonymous

    Thank you for your explanation.
    It's very detailled and clear, and with examples.

  15. Anonymous

    Thank you very much ! Save my day !

  16. Anonymous

    Thank you so much for sharing this information!

    When I tried to use resize2fs I got a error:

    "Bad magic number in super-block while trying to open..."

    I think resize2fs does not work with XFS partitions, so I had to use lvresize with -r option to "resize underlying filesystem together with the logical volume using fsadm":

    sudo lvresize -rl +100%FREE /dev/centos/root

    Hope it helps someone, and thanks again!

  17. Anonymous

    Thanks for the information. It helped me out and worked just fine! Amazingly, I was unable to find the process documented anywhere else! Thanks again.

     

  18. Anonymous

    how to create a virtual machine through installation CD plugged into NFS server, on to a Xenserver on pre-existing NFS VHD. we are using citrix xencenter.