CAM/SCSI Errors - Can I swap these cables?

[sce28335]

CAM/SCSI Errors, JBOD Cabling, and Stability Concerns

Hi all,

I’m working through recurring instability (CAM errors, checksum errors under load) on a large pool and want to confirm if my cabling layout is the root issue.

Hardware

• Host HBA: Broadcom MegaRAID 9480-8i8e
• JBOD #1: Supermicro 846BE1C-R1K03JBOD
• JBOD #2: Supermicro 946SE1C-R1K66JBOD
• TrueNAS: 12.0-U7

Issue

Pool imports (after many errors while syncing), all disks enumerate, but under heavy reads/writes:

• CAM status errors appear
• Checksum errors rise
• Scrubs “heal” the errors, but they reoccur on the next I/O burst

Vendor Recommendation

Support told me the intended layout is:

• Host → JBOD1 TOP (IN)
• JBOD1 BOTTOM (OUT) → JBOD2 TOP (IN)

The Question

• Is my current setup (host on bottom, JBOD2 on top) the cause of these pathing issues?
• Does the expander truly require host on the labeled IN row and expansion on IN/OUT, or are the ports fully symmetric?
• If I shut down, recable as per vendor instructions, and reboot, should TrueNAS/ZFS see all disks normally (since drives are tracked by gptid/GUID)?

Plan if change is needed

1. Export pool (zpool export Archive) for safety
2. Shutdown host → JBOD2 → JBOD1
3. Re-cable: Host → JBOD1 TOP (IN), JBOD1 BOTTOM (IN/OUT) → JBOD2 (IN)
4. Power up in order: JBOD1 → JBOD2 → Host
5. Import pool (should happen automatically), verify zpool status

Does this plan follow the correct sequence?

Would appreciate confirmation from others running similar Supermicro JBOD stacks.

Thanks!

[NugentS]

What firmware are you using in that MegaRAID card? Is it IT mode or IR?
You should be using an HBA, not a RAID Card

[sce28335]

Is there a way to verify this without shutting down the system?

[sce28335]

The card is in IT mode, we have been running this system for a few years and the disks are listed by their identifiers.

Here are photos of the cabling/diagram sent of a similar system.