Host Storage

Ondat uses the storage available on the nodes where it is installed to present as available for volumes.

To mitigate against problems caused by filling the host root disk, we recommend mounting a separate device into the /var/lib/storageos directory. Ondat is agnostic to the type of filesystem mounted in /var/lib/storageos.

Extending Available Storage

Ondat uses subdirectories of /var/lib/storageos/data to hold user data. By default, the directory /var/lib/storageos/data/dev1 will be created when a node is bootstrapped, and used for pool data. It is possible to shard the data by creating more directories into this structure. Ondat will save data in any directory that conforms to the pattern /var/lib/storageos/data/dev[0-9]+, such as /var/lib/storageos/data/dev2 or /var/lib/storageos/data/dev5. This functionality enables operators to mount different devices into devX directories and Ondat will recognise them as available storage automatically.

There are two possible options to expand the available disk space for Ondat to allocate:

1. Mount filesystem in /var/lib/storageos/data/devX
2. Use LVM to expand the logical volume available to Ondat

Option 1: Mount Additional Devices

This option enables operators to expand the cluster’s available space at any time without having to stop applications or forcing operational downtime. The expansion of disk is transparent for applications and Ondat Volumes. Ondat will use the new available space to create new data files.

1. Context

   We assume that there is a disk available in our Linux system without formatting in addition to the root filesystem. Ondat data dir dev1 (/var/lib/storageos/data/dev1) is using /dev/xvda1. We will use the device /dev/xvdf to expand Ondat available space.

   List available block devices in the host.

   root@node0:~# lsblk
   NAME    MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
   xvda    202:0    0  128G  0 disk
   `-xvda1 202:1    0  128G  0 part /
   xvdf    202:80   0  100G  0 disk
   

   Check Ondat cluster’s available capacity.

   $ storageos get node -ojson | jq -r '.[] | { name: .name, capacity: .capacity.total  }'
   "node0"
   137,438,953,472
   "node1"
   137,438,953,472
   "node2"
   137,438,953,472
   

2. Format device

   root@node0:/var/lib/storageos/data# mkfs -t ext4 /dev/xvdf
   mke2fs 1.42.12 (29-Aug-2014)
   Creating filesystem with 26214400 4k blocks and 6553600 inodes
   Filesystem UUID: 380712fa-6f82-477a-81a5-d7466d4c6b7f
   Superblock backups stored on blocks:
           32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632, 2654208,
           4096000, 7962624, 11239424, 20480000, 23887872
   
   Allocating group tables: done
   Writing inode tables: done
   Creating journal (32768 blocks): done
   Writing superblocks and filesystem accounting information: done
   

3. Mount filesystem

   root@node0:~# mkdir -p /var/lib/storageos/data/dev2
   root@node0:~# mount /dev/xvdf /var/lib/storageos/data/dev2
   

4. Verify available storage

   In less than 30 seconds, Ondat will see the new available capacity.

   $ storageos get node -ojson | jq -r '.[] | { name: .name, capacity: .capacity.total  }'
   "node0"
   244,491,013,324
   "node1"
   137,438,953,472
   "node2"
   137,438,953,472
   

   Note that the node node0 has increased the TOTAL capacity in 100Gi.

💡 Persist the mount at boot by adding the mount endpoint to /etc/fstab

Option 2: Expand Existing Devices Backed by LVM

This option enables operators to take advantage of LVM to manage disks.

1. Context

   We assume that /var/lib/storageos is mounted onto an LVM volume. We are using a volumegroup named storageos and logical volume called data. There is a second physical disk /dev/xvdg unused.

   List available block devices in the host.

   root@node2:~# lsblk
   NAME             MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
   xvda             202:0    0  128G  0 disk
   `-xvda1          202:1    0  128G  0 part /
   xvdf             202:80   0  100G  0 disk
   `-storageos-data 254:0    0   99G  0 lvm  /var/lib/storageos
   xvdg             202:96   0  100G  0 disk
   

   Check Ondat cluster’s available capacity.

   $ storageos get node -ojson | jq -r '.[] | { name: .name, capacity: .capacity.total  }'
   "node0"
   137,438,953,472
   "node1"
   137,438,953,472
   "node2"
   107,696,304,947 # --> LVM storageos/data volume
   

2. Add physical disk to LVM

   root@node2:~# vgextend storageos /dev/xvdg
     Volume group "storageos" successfully extended
   

   The volume group storageos must have 2 physical volumes (#PV)

   root@node2:~# vgs
     VG        #PV #LV #SN Attr   VSize   VFree
     storageos   2   1   0 wz--n- 199.99g 104.99g
   

3. Extend logical volume data

   root@node2:~# lvextend -L+100G /dev/storageos/data
     Size of logical volume storageos/data changed from 95.00 GiB (24320 extents) to 195.00 GiB (49920 extents).
     Logical volume data successfully resized
   

4. Resize the FileSystem

   ⚠️ Your filesystem must support the option to be expanded, and to do so while in use. Otherwise, you need to unmount first.

   root@node2:~# resize2fs /dev/storageos/data
   resize2fs 1.42.12 (29-Aug-2014)
   Filesystem at /dev/storageos/data is mounted on /var/lib/storageos; on-line resizing required
   old_desc_blocks = 6, new_desc_blocks = 13
   The filesystem on /dev/storageos/data is now 51118080 (4k) blocks long.
   

5. Check new available space

   The mounted file system to /var/lib/storageos has increased its size.

   root@node2:~# df -h /dev/mapper/storageos-data
   Filesystem                  Size  Used Avail Use% Mounted on
   /dev/mapper/storageos-data  192G   60M  183G   1% /var/lib/storageos
   

   Ondat available storage has increased too.

   $ storageos get node -ojson | jq -r '.[] | { name: .name, capacity: .capacity.total  }'
   "node0"
   137,438,953,472
   "node1"
   137,438,953,472
   "node2"
   206,158,430,208 # --> 100G more available
   

💡 Persist the mount at boot by adding the mount point to /etc/fstab