Expand GFS2 datastore

[rasmusan]

Hello,

I have a GFS2 datastore that I have expanded on the FC storage array. On the hosts when running "sudo multipath -ll" the new size is not reflected. Tried multiple ways to rescan etc., but only after a reboot of one host, that particular host reflects the new size when running "sudo multipath -ll"

How should I proceed to get the other hosts in the cluster to identify new LUN size, without rebooting?

What is the correct way to expand a GFS2 datastore end to end on HVM cluster?

[CalvinZito]

I know this was answered before here - here's one of the times it was answered: https://community.hpe.com/t5/hpe-morpheus-vm-essentials/morpheus-shows-incorrect-capacity-after-extending-gfs2-datastore/m-p/7259940#M1600. 

I'd suggest using the search at the top, change it "users". Once you get the results, at the top, look at Forums, and on the left update  to look in Software > HPE Morpheus Software > HPE Morpheus VM Essentials. That's where I found this response. There might be others that would help.

[Arnout_Verbeken · Accepted solution]

Increasing the Size of a GFS2 Filesystem on an iSCSI LUN (HVM Cluster)

This document describes the procedure to increase the size of a GFS2 filesystem backed by an iSCSI LUN in an HVM cluster.

Important assumptions

• The backing iSCSI LUN has already been expanded on the storage array.

• The LUN is presented to all HVM cluster nodes via multipath.

• The GFS2 filesystem is already mounted on all nodes (shared cluster filesystem).

---

Overview

The high‑level steps are:

1. Rescan the SCSI devices on all cluster nodes so they see the new LUN size.

2. Resize the multipath device on all cluster nodes.

3. On one node only, grow the GFS2 filesystem.

4. On the remaining nodes (or all nodes), trigger a GFS2 statfs sync.

5. Allow HVM Manager / Morpheus to detect and display the new datastore size.

---

Step 0 – Identify Devices and Mount Points

Before starting, on one host gather:

• The underlying SCSI device(s) (e.g. sdb, sdc, …) behind the multipath device.

• The multipath WWID (e.g. 36000XXX0X0000000000000XXXXX0db).

• The GFS2 mount point (e.g. /mnt/ffaXXXXX-aXXX-XXX0-XXX6-XXXXX2acd9e).

• The GFS2 cluster and datastore names (for step 4).

Example commands:

# Show multipath devices and WWIDs 
multipath -ll 
# Show GFS2 filesystems and mount points 
mount | grep gfs2 
# Show GFS2 cluster and datastore names 
ls /sys/fs/gfs2​

You should see something like:

• Multipath WWID: 36000XXX0X0000000000000XXXXX0db

• Mount point: /mnt/ffaXXXXX-aXXX-XXX0-XXX6-XXXXX2acd9e

• GFS2 name: <clustername>:<datastorename>

Use your actual values in the steps below.

---

Step 1 – Rescan the Underlying SCSI Devices (All Hosts)

On every HVM cluster node, rescan the SCSI device(s) backing the multipath LUN so the OS sees the new size.

For each relevant sdX device (e.g. sdb, sdc):

echo 1 > /sys/block/sdX/device/rescan

Example:

echo 1 > /sys/block/sdb/device/rescan

• Repeat for all underlying paths (all sdX) used by the multipath device.

• Run this on all cluster hosts.

• Confirm the new size by running:

• lsblk /dev/sdX

---

Step 2 – Resize the Multipath Map (All Hosts)

Next, on every HVM cluster node, instruct multipathd to resize the multipath device to use the new LUN size.

multipathd -k'resize map 36000XXX0X0000000000000XXXXX0db'

Replace 36000XXX0X0000000000000XXXXX0db with your actual LUN WWID from Step 0.

• Run this command on all hosts that participate in the cluster.

• This updates the size of the multipath device to match the expanded LUN.

---

Step 3 – Grow the GFS2 Filesystem (One Host Only)

Once all hosts see the new multipath size, perform the GFS2 filesystem grow operation from one cluster node only.

On a single host:

gfs2_grow -v /mnt/ffaXXXXX-aXXX-XXX0-XXX6-XXXXX2acd9e

Where /mnt/ffaXXXXX-aXXX-XXX0-XXX6-XXXXX2acd9e is the GFS2 mount point obtained in Step 0.

Notes:

• gfs2_grow is online and non‑destructive.

• Ensure the filesystem is mounted and the cluster is healthy before running.

---

Step 4 – Trigger GFS2 Statfs Sync (Other Hosts / All Hosts)

To ensure all nodes have up‑to‑date free space information, trigger a statfs sync on the remaining hosts.
This can be safely run on all hosts, including the one where you ran gfs2_grow, as it is non‑destructive.

On each host:

echo 1 > '/sys/fs/gfs2/<clustername>:<datastorename>/statfs_sync'

Replace <clustername>:<datastorename> with the actual GFS2 name (e.g. hvmcluster01:ds01).

You can list available GFS2 names with:

ls /sys/fs/gfs2

---

Step 5 – HVM Manager / Morpheus UI Update

No manual UI action is required.

• The HVM Manager / Morpheus platform will automatically detect the new size of the datastore on its next sync cycle.

• After the next sync, the datastore’s new capacity will be reflected in the UI.

If you need to confirm immediately, you can:

• Check filesystem size on a host:

  df -h /mnt/ffaXXXXX-aXXX-XXX0-XXX6-XXXXX2acd9e

• Then validate that HVM Manager / Morpheus shows the updated size once it has synced.

---

Summary

• Ensure the LUN is expanded on the storage array (pre‑requisite).

• Rescan SCSI devices and resize the multipath map on all cluster nodes.

• Run gfs2_grow once from a single node.

• Trigger statfs_sync on the remaining (or all) nodes.

• Let HVM Manager / Morpheus automatically pick up and display the new size.

[MaartenV1]

Is this something that will be more friendly in next versions like "increase Datastore" within VMware is ?