Skip to main content
Version: 2604 (Preview)

Fibre Channel Fabric Maintenance

Reference Azure

DOCUMENT CATEGORY: Reference SCOPE: Fibre Channel fabric maintenance PURPOSE: Reference for FC switch firmware update, port replacement, fabric merge/split MASTER REFERENCE: Microsoft Learn — Enable External Storage

Status: Active

This document provides maintenance procedures for Fibre Channel fabrics used in Azure Local SAN-attached deployments, covering firmware updates, port and SFP replacement, and a recurring maintenance schedule. Procedures cover both Brocade Fabric OS and Cisco MDS NX-OS platforms.

FC Switch Firmware Update​

Firmware updates require a maintenance window — typically one fabric at a time so the cluster remains operational on the surviving fabric. MPIO on each node automatically routes I/O through the active fabric while the other is updated.

Pre-Update Checklist​

  1. Confirm current MPIO path count on each node: 
    mpclaim -s -d
  2. Confirm the cluster is Healthy: 
    Get-Cluster
    Get-ClusterNode
  3. Schedule maintenance window (~2 hours per fabric).
  4. Notify relevant stakeholders and create a change record.

Brocade Firmware Update​

# SSH to the switch
ssh admin@brocade-fab-a.corp.azurelocal.cloud

# Check current version
firmwareshow

# Stage firmware (FTP from staging server)
firmwaredownload

# Switch reboots automatically — wait ~10 minutes
# After reboot, verify
firmwareshow

During the update, Fabric B continues serving I/O. After Fabric A returns, MPIO automatically reclaims paths within ~30 seconds.

Cisco MDS Firmware Update​

install all kickstart bootflash:m9700-sf3ek9-kickstart-mz.X.Y.Z.bin \
            system    bootflash:m9700-sf3ek9-mz.X.Y.Z.bin

For a non-disruptive upgrade (NDU), confirm both supervisors are present and active before initiating the install.

Post-Update Validation​

mpclaim -s -d
Get-PhysicalDisk | Where-Object BusType -eq 'Fibre Channel' |
    Select-Object FriendlyName, OperationalStatus

Path count must match the pre-update value. If paths do not recover within 60 seconds, check ISL connectivity and zone enforcement on the updated switch before escalating.

Port (Cable/SFP) Replacement​

When a fabric port shows persistent CRC errors or link flaps, use the following procedure to isolate and replace the affected component with minimal I/O disruption.

  1. Identify the affected port:
    • Brocade: porterrshow
    • Cisco MDS: show interface counters
  2. Cross-reference WWPN aliases to identify the affected node and HBA port.
  3. Create a change record.
  4. Disable the affected port to force MPIO failover:
    • Brocade: portdisable <port>
    • Cisco MDS: enter interface fc1/N then shutdown
  5. Wait 60 seconds and confirm MPIO has failed all paths for that port over to the redundant fabric.
  6. Replace the cable and/or SFP.
  7. Re-enable the port; wait 30 seconds; confirm path count recovers to the expected value.
  8. Reset error counters to establish a clean baseline:
    • Brocade: portstatsclear <port>
    • Cisco MDS: clear counters interface fc1/N
Cleaning fiber connectors

Use an inspection microscope and end-face cleaner before inserting replacement SFPs. Dirty connectors are a leading cause of CRC errors on single-mode links.

Recurring Maintenance Schedule​

TaskFrequencyOwner
Review fabric error countersWeeklyOperations
Confirm zoneset matches designMonthlyStorage admin
Firmware updatePer vendor advisory or annuallyStorage admin
Cable inspection / fiber cleaningAnnuallyOn-site technician

References​

Cross-references​

  • Phase 03 Task 13 — Configure MPIO & MSDSM — MPIO verification commands referenced in the firmware update and port replacement sections above.
  • Appendix P — FC Fabric Configuration — fabric zoning and port layout that this appendix maintains.

Procedures verified against Brocade Fabric OS 9.x and Cisco MDS NX-OS 9.x as of 2026-05-02.

For questions about these procedures, refer to the vendor documentation linked above or contact the Azure Local Cloud documentation team.