Commands
Deploy resilient, scalable on-prem Kubernetes storage using Rook-Ceph. Learn architecture, setup steps, best practices, and examples for block, file, and object storage. Table of Contents
Learn how to configure Multi-Path I/O (MPIO) for SAN-based storage on RHEL 9 and CentOS 9. Step-by-step guide with CLI examples, best practices, and troubleshooting tips.
Enterprise Linux environments running RHEL 9 or CentOS Stream 9 frequently rely on SAN-based storage for performance, scalability, and resilience. To fully benefit from redundant storage paths provided by Fibre Channel (FC) or iSCSI SANs, Multi-Path I/O (MPIO) must be correctly configured.
This guide walks through a production-ready, search-optimized, and vendor-agnostic approach to configuring MPIO on RHEL 9 and CentOS 9. It includes clear explanations, CLI examples, tables, and best practices to help system administrators build a reliable storage stack.
Multi-Path I/O allows a Linux host to access the same storage LUN through multiple physical paths. Instead of the OS seeing multiple block devices, MPIO presents a single logical device and manages path selection, failover, and load balancing.
🟢 Key Benefits of MPIO |
|
|
|
|
RHEL 9 and CentOS 9 support MPIO across common SAN transports:
| Protocol | Supported | Notes |
|---|---|---|
| Fibre Channel (FC) | Yes | Requires HBA and zoning |
| iSCSI | Yes | Works with software and hardware initiators |
| FCoE | Yes | Requires DCB-enabled NICs |
| NVMe-oF | Partial | Uses native multipath (not dm-multipath) |
This article focuses on dm-multipath, which applies to FC and iSCSI storage.
Without multipathing:
/dev/sdb /dev/sdc /dev/sdd
With multipathing:
/dev/mapper/mpatha → /dev/sd[bcd]
The device-mapper multipath layer aggregates multiple block devices into one resilient logical volume.
🔧 Prerequisites |
Before configuring MPIO, verify the following:
| Requirement | Description |
|---|---|
| OS | RHEL 9 or CentOS Stream 9 |
| SAN Connectivity | FC or iSCSI paths visible |
| Root Access | Required for configuration |
| LUNs | Presented from SAN (do not format yet) |
Confirm that LUNs are visible:
lsblk
Install the required utilities:
sudo dnf install -y device-mapper-multipath device-mapper-multipath-libs
Verify installation:
rpm -qa | grep multipath
Enable multipathing using the built-in configuration helper:
sudo mpathconf --enable --with_multipathd y
This:
|
|
|
Start and enable the service:
sudo systemctl enable --now multipathd
Check status:
systemctl status multipathd
The default RHEL 9 configuration includes:
defaults {
find_multipaths yes
user_friendly_names no
}
🟢 Why These Defaults Matter |
| Setting | Reason |
|---|---|
find_multipaths yes | Prevents multipath on single-path devices |
user_friendly_names no | Ensures stable WWID-based naming |
💡Best Practice: Keep |
List detected multipath devices:
sudo multipath -ll
Example output:
mpatha (3600508b400105e210000900000490000)
size=500G features='1 queue_if_no_path'
|-+- policy='service-time 0' prio=50 status=active
| |- sdb ready running
| |- sdc ready running
🔑 Key Fields Explained |
| Field | Meaning |
|---|---|
| WWID | Unique SAN identifier |
| Policy | Load balancing method |
| Prio | Path priority |
| Status | Active/ready paths |
/etc/multipath.confFor enterprise SANs (NetApp, Dell EMC, HPE, IBM), vendor-specific tuning is recommended.
🖥️ Example Production Configuration |
defaults {
find_multipaths yes
polling_interval 5
path_selector "service-time 0"
path_grouping_policy group_by_prio
path_checker tur
prio alua
failback immediate
no_path_retry queue
features "1 queue_if_no_path"
}
blacklist {
devnode "^sd[a]$"
}
Reload configuration:
sudo systemctl reload multipathd
Re-scan paths:
sudo multipath -r
Use the following commands to validate multipathing:
multipath -ll
lsblk -o NAME,KNAME,TYPE,SIZE
Expected result:
|
|
🟢 Using LVM (Recommended) |
pvcreate /dev/mapper/mpatha
vgcreate vg_san /dev/mapper/mpatha
lvcreate -n lv_data -L 400G vg_san
mkfs.xfs /dev/vg_san/lv_data
Mount persistently:
mkdir /data
mount /dev/vg_san/lv_data /data
Add to /etc/fstab:
/dev/vg_san/lv_data /data xfs defaults,_netdev 0 0
Simulate a path failure:
|
Observe behavior:
multipath -ll
Expected outcome:
|
|
|
| Command | Purpose |
|---|---|
multipath -ll | View path status |
multipath -v3 | Debug device discovery |
journalctl -u multipathd | Service logs |
ls -l /dev/disk/by-id/ | Validate WWIDs |
|
|
|
|
|
|
|
|
|
Deploy resilient, scalable on-prem Kubernetes storage using Rook-Ceph. Learn architecture, setup steps, best practices, and examples for block, file, and object storage. Table of Contents
Self-explanatory placeholder (typically) matches the keyword catch phrase statement. Table of Contents 🔈Introduction Migrating from Ext4 to Btrfs has become a common objective for Linux
Learn how to configure LVM on multipath devices in RHEL with this comprehensive step-by-step guide. Improve storage performance, ensure high availability, and master enterprise-level disk