macOS 15.5 + SANLink3 F2 + Xsan: FC Write Timeout and c3timeout Errors

  • 28 Views
  • Last Post 21 hours ago
terry wang posted this 22 hours ago

Hi all,

 

We’ve recently encountered a serious Fibre Channel I/O timeout issue when using macOS 15.5 with a Promise SANLink3 F2 adapter in an Xsan environment.

 

### Environment:

- macOS Sequoia 15.5 (latest public release)

- Promise SANLink3 F2 (dual-port 16Gb FC)

- Apple Xsan with MDC + metadata LUNs + data LUNs on FC storage

- Brocade FC switch (with zoning configured)

- Multipath (MPIO) enabled

 

### Problem Description:

On macOS 15.5 systems connected to Xsan via SANLink3 F2, we consistently observe **write I/O timeouts**, followed by **abort and chip timeout errors**. These show up in the system log as:

 

This corresponds to failed write operations (SCSI op 0x2A) and eventually causes the mount point to hang or file system-level errors in Xsan volumes.

 

### Diagnostics Performed:

- `log show --last 5m | grep sanlink` confirms repeated timeouts.

- FC switch shows **extremely high c3timeout counts** on port 32 (~89,900).

- However, port 32 is **not physically connected** or zoned — it’s unused.

- No CRC, sync loss, or PCS errors were found on other active FC ports (32–47).

- Disabling port 32 on the switch improves system behavior temporarily.

- Suspected residual MPIO paths pointing to nonexistent targets may be involved.

 

### Attempts to Resolve:

- Re-seated SFPs and fibre cables

- Reinstalled latest Promise SANLink drivers (still no official macOS 15.5 support)

- Disabled unused FC ports at the switch level

- Verified zoning is correct for active LUNs

- Tried removing/rebuilding volume metadata in Xsan

 

### Key Question:

Has anyone else seen **write timeouts or c3timeout-related errors** with macOS 15.5 and SANLink3 F2 in an Xsan setup? Could this be due to compatibility issues with SANLink drivers on macOS 15.5?

 

Any insight or suggestions would be much appreciated!

 

---

 

Thanks in advance,

 

Order By: Standard | Latest | Votes
R P posted this 21 hours ago

Hi Terry,

- FC switch shows **extremely high c3timeout counts** on port 32 (~89,900).
- However, port 32 is **not physically connected** or zoned — it’s unused.

These are errors on the Switch, and on an unconnected port, I'm not seeing any connection between this error and the SanLink.

If the errors were on the port(s) connected to the SanLink, that would be one thing, but that is not the case.

Is there something I'm missing?

terry wang posted this 21 hours ago

Port 32 was previously connected to one of the controllers on our SAN storage system. However, that particular storage controller had failed, and although the Fibre Channel cable was still physically connected, the port continued to report errors.

I’ve just removed the fibre cable from port 32, so we expect that specific issue to be resolved now.

 

However, the core problem still persists — when using AJA System Test Lite to test read/write performance, we consistently encounter the following kernel errors:

kernel: (SANLink-FC) [SL] driver timeout tgt:17 lun:2 ctrl:2 op 2a ...

kernel: (SANLink-FC) [SL] abort xxxx

 

kernel: (SANLink-FC) [SL] chip timeout ...

At the same time, AJA System Test Lite halts completely during the test.

Search
Categories
Close