SAN Fabric Path Redundancy Lost – Quick Fix

You're not the first to see this

It's a Monday morning and your SAN monitoring lights up orange. Paths that were active last week are gone. You've got a fabric with no redundancy. Annoying, right? I've fixed this exact mess a dozen times. Here's what you do.

The Fix – Step by Step

Step 1: Check your optics and cables

Don't bother with software first. Walk to the switch. Look at the LEDs. If a port shows amber or no light, swap the SFPs and cables. I've seen brand-new SFPs from the same batch fail. Pull one, replace it with a known good one. Verify the cable isn't bent too tight. This fixes about 40% of cases within 5 minutes.

switch# show interface fc2/1

Run that on your Cisco MDS or Brocade. Look for up or down. If it's down, check physical layer first.

Step 2: Check zoning

Most lost redundancy comes from a zone that got merged wrong. Someone added a new server port and accidentally removed an old one. Or two zones conflict. I've seen this happen after a SAN admin left a zone config half-done.

switch# show zoneset active

Compare the active zone set to your baseline config. If you don't have a baseline, that's your problem. Every path should be in at least two zones – one for each fabric. If you see a host WWN in only one zone, that's your missing path.

Step 3: Verify ISL links

If zones look fine, check inter-switch links (ISLs). One dead ISL can break redundant paths between fabrics. Run:

switch# show isl status

On Brocade, that shows the ISL table. Any link down? That kills path redundancy for all devices behind it. Fix that link or add another ISL.

Why This Happens

Redundancy relies on two separate paths from server to storage through two fabrics. Each fabric must have its own zone set, its own ISL links, and its own optics. The common failure points are:

Zoning misconfig – A zone entry gets deleted or a WWN gets duplicated across fabrics.
Optics failure – SFPs die slowly or suddenly. They're the weakest link.
Cable damage – A patch cable gets crushed behind a rack. It still passes light but at degraded levels.
ISL congestion – Rare, but a flapping ISL can disable paths.

Less Common Variations

Firmware bugs

I once had a Brocade FOS 7.4.1a that dropped ISL paths after a firmware upgrade. The fix was upgrading to 7.4.1b. Check your switch firmware release notes for known issues. Sometimes a patch resolves phantom path loss.

Trunking issues

If you use trunking on Brocade or Cisco, one member port failing can drop the whole trunk. Run:

switch# show trunk

If you see a trunk with fewer ports than expected, that's your problem. The fix is usually reseating the SFPs on those ports or re-creating the trunk config.

Fabric binding

Some shops use fabric binding to restrict which switches can join. If a new switch was added without updating the binding list, it gets isolated. That breaks path redundancy for hosts connected to that switch. Check with:

switch# show fabric-binding

If it's empty, binding isn't the issue. If it's populated and a switch WWN is missing, add it.

Prevention – Don't Let It Happen Again

Three things to set up now:

Document your zoning baseline. Save the active zone set after any change. Compare it monthly. I use a simple script that diffs the config against a stored copy.
Monitor optics health. Most SAN switches report SFP diagnostics. Run show interface transceiver details on Cisco or sfpshow on Brocade. Look for high bit error rates (BER) above 1e-15. Replace any SFP with BER over 1e-12 – it's about to fail.
Test path failover quarterly. Shut one ISL or one zone intentionally. See if your storage multipath software (like PowerPath or native MPIO) switches to the other path. If it doesn't, you've got a config bug that you can fix before it breaks in production.

That's it. You're back to redundant paths in under 30 minutes. I've done this faster with a spare SFP in hand. Keep a few on the shelf – they cost peanuts compared to downtime.