GlusterFS settings

GlusterFS is a scalable, distributed file system that aggregates disk storage resources from multiple servers into a single global namespace. GlusterFS features high expandability, high availability, and super-high performance, and it avoids single point of failures because it does not have any metadata servers. ARM host do not support GlusterFS.

Application scenarios

GlusterFS is applicable to school scenarios that require high system performance.

Configuration flow

Add a vSwitch

For a cluster to use GlusterFS storage, you must add a storage vSwitch with the same name on each host that has GlusterFS configured. For information about vSwitches, see "Manage virtual switches for a host."

If the management node in the cluster is not used as a service node, to use GlusterFS storage, you must add a storage vSwitch on the management node, and make sure the name of the storage vSwitch is the same on each node. Add the management node to the cluster, add a storage vSwitch, and then delete the management node from the cluster.
In the stateful failover scenario, no matter whether or not the backup node joins the GlusterFS cluster, a storage vSwitch must be added, and its name must match other nodes to ensure proper GlusterFS storage usage during failover. If the standby node does not have a storage vSwitch with the same name as the active node, but an active/standby switchover is performed, restore GlusterFS settings. For more information about how to restore GlusterFS settings, see H3C Workspace Cloud Desktop Maintenance Guide.

Teaching storage initialization
This task allows you to format an idle disk on a host and mount it to the local storage path /vms/learningspace (For more information about the procedures, see "Teaching storage initialization."). This task also allows you to mount the shared storage in the cluster to the local storage path /vms/learningspace (For more information about the procedures, see "Teaching storage initialization."). The newly created GlusterFS mounting path and storage block will be mounted to the /vms/learningspace/glusterfs path on the same disk or the shared storage.
Create a GlusterFS volume

Create a GlusterFS volume

To create a volume of the custom or stateful failover type, follow these restrictions and guidelines:

For a storage block to use the space on a drive of the host, make sure the storage block directory starts with the directory for mounting the drive on the host. In addition, the mount point and the storage block directory cannot start with /vms/. For example, if the drive is mounted on /test/brick, use /test/brick/XX as the storage block directory and /test/brick/gluster as the directory for mounting the GFS file system as a best practice.
Make sure the directory where the GFS file system will be mounted is not a sub directory of the directory where the storage block is stored, and vice versa. For example, if the directory for mounting the GFS file system is /test/brick/gluster, the directory for storing the storage block cannot be /test, /test/brick, /test/brick/gluster, or /test/brick/gluster/XX.
To avoid mounting information loss after a server reboot when you use an additional disk, mount and format the disk in the back end and then write mounting information into the /etc/fstab file.
As a best practice, use SSD hard drives and virtual switches with 10GbE NICs.

Procedure

From the left navigation pane, select Data Center > Teaching Storage > GlusterFS Settings.
Click Create.
Configure the parameters as required, and then click OK.

Parameters

Type: Select a storage type. If shared storage is created during the initialization of the teaching storage, you cannot create GlusterFS storage of the Teaching Image Storage type. If you select Teaching Image Storage, the mount point is /vms/learningspace/glusterfs/courseImages and the storage block path is /vms/learningspace/glusterfs/brick. If you select Custom Storage, you can define a mount point and storage block path as needed. If you select Stateful Failover Storage, GFS storage is used as the shared storage for the stateful failover system, but the mount point and storage block path cannot begin with /vms/. For the Stateful Failover Storage type, you can define a mount point and storage block path as needed, but the mount point and storage block path cannot begin with /vms/. A GFS volume of the Custom Storage or Stateful Failover Storage type cannot be used as the course storage path.
Host Pool: Select a host pool.
Storage Cluster: Select a cluster.
Host: Select a host for storing course images. The number of hosts must be a multiple of the number of backups in the topology type.
vSwitch: Select the vSwitch to which the host attached. The selected virtual switch must be configured with an IP address. For example, when the storage network and external network use the same physical interface, they will use different subnets on the same virtual switch vs_storage. However, this virtual switch does not have an IP address, and you cannot select this virtual switch when creating new GlusterFS storage.
Topology Type: Select the number of GlusterFS storage backups. You can select 2 to 10 backups depending on the cluster scale and backup requirements.
GFS Mount Point: Directory where the GFS storage will be mounted.
Storage Block Path: Directory where the GlusterFS physical storage block resides.

Manage GlusterFS volumes

Expand a GlusterFS volume

If you use a host that has acted as a GlusterFS node for storage expansion, make sure the bricks on the original GlusterFS volume have been deleted.
Before expansion, create and mount the same storage block directory on a newly-added host for expansion as the existing storage block path. For a storage block to use the space on a drive of the host, make sure the storage block directory starts with the directory for mounting the drive on the host. For more information, see "Create a GlusterFS volume."

Perform this task to add storage backups or add hosts that acts as storage nodes to expand a GlusterFS volume.

From the left navigation pane, select Data Center > Teaching Storage > GlusterFS Settings.
Click Expand in the Actions column for a GlusterFS volume.
In the dialog box that opens, select whether to change the topology type. If you select to not change the topology type, make sure the number of available hosts in the cluster is a multiple of the number of backups in the topology type.
Select target hosts, and then click OK.

Shrink a GlusterFS volume

Prerequisites

This task might result in severe or undesirable impact. Back up the data in the desired GlusterFS volume.

Restrictions and guidelines

As a best practice, use two replicas.

For better storage performance, do not use the full replica policy when over three hosts exist.

Procedure

From the left navigation pane, select Data Center > Teaching Storage > GlusterFS Settings.
Click Shrink in the Actions column for a GlusterFS volume.
In the dialog box that opens, select a host, select whether to delete the storage pool, select I am fully aware of the impacts of this operation, enter the password of the current administrator, and then click OK.

Delete a GlusterFS volume

From the left navigation pane, select Data Center > Teaching Storage > GlusterFS Settings.
Click Delete in the Actions column for a GlusterFS volume.
In the dialog box that opens, click OK.

View GlusterFS storage information

From the left navigation pane, select Data Center > Teaching Storage > GlusterFS Settings. On the Overview tab, you can view the topology, storage node states, and partition states.

Repair a storage node

From the left navigation pane, select Data Center > Teaching Storage > GlusterFS Settings.
Click the Node Management tab.
Click Repair Node in the Actions column for a node that has mounting or partition connection errors.

If the GlusterFS storage pool is inactive, and cannot be manually started, perform the following steps to resolve the issue:

‍Verify that the storage network is reachable.
Mount the GlusterFS volume to a temporary directory to check for duplicate files:
1. ‍Execute the mkdir -p /vms/tmp command to create temporary directory /vms/tmp.
2. Execute the mount -t glusterfs IP:/VolumeName /vms/tmp command to mount the GlusterFS volume. IP represents the IP address of a node on the GlusterFS volume, and VolumeName represents the name of the volume.
3. Execute the df -h command to verify that the volume has been mounted successfully.
4. Execute the cd /vms/tmp command to access the directory, and execute the ls -a command to check for duplicate files. Back up duplicate files and execute the rm -r duplicate file name command to delete the duplicate files, if any.
Execute the gluster volume heal VolumeName info command to check for the ls in split brain file. Delete the file, if any. To reserve the file, save the file in another directory, and then copy the file to its original directory after the storage pool is recovered.