If a brand-new RAID5 alternates between one missing disk and all disks missing, stop rebuilding and verify that all three drives are detected consistently first. That pattern points more strongly to power, cabling, backplane/controller, or disk-detection instability than to three simultaneous RAID failures.
Current ZimaOS RAID5 requires at least three disks and should show a failed member as Degraded. Repeatedly rebooting does not fix an intermittently disappearing drive path.
Step 1: Check Every Disk Outside the RAID View
lsblk -o NAME,SIZE,MODEL,SERIAL,TRAN
dmesg | grep -Ei 'ata|sas|scsi|reset|error|fail' | tail -150Run the checks after several cold boots. All three 16TB disks should appear every time with stable model/serial information.
Step 2: Reseat Power and Data Connections
Power down fully before reseating drives, SATA/SAS cables, backplane connectors, or HBA connections. A large 3.5-inch disk can disappear during spin-up when power is marginal.
Step 3: Check SMART on Each Drive
smartctl -x /dev/sdXLook for failed self-tests, media errors, command timeouts, pending sectors, or interface errors. A new disk can still be defective.
Step 4: Inspect RAID State
The current ZimaOS storage setup guide says a failed member should make the array Degraded while data remains accessible.
Advanced users can also inspect:
cat /proc/mdstat
mdadm --detail /dev/mdXUse these only if the current array is md-backed and you know the correct device.
Do Not Recreate RAID Until Detection Is Stable
If all disks vanish intermittently, a clean RAID creation will simply fail again. Fix the hardware/controller path first.
RAID5 Requires Three Reliable Members
The current ZimaOS RAID5 guide describes RAID5 as single-parity storage that survives one member failure. It is not designed to tolerate unstable detection of multiple drives.
Back Up Before Rebuilding an Existing Array
If the pool already contains data, do not delete/recreate it merely because the UI reports a missing member. Protect data first and identify whether the drive truly failed or only disappeared from the controller.
Check Replacement/Rebuild Logic Separately
If one specific disk is consistently failed, replace only that disk and follow ZimaOS's rebuild workflow. Do not initialize the replacement randomly or add it to a new pool.
The RAID rebuild troubleshooting guide provides the next branch after detection is stable.
FAQ
Why does a fresh RAID5 say Missing Disk?
A disk may have dropped out during creation or may not be detected reliably by power, cabling, backplane, or controller.
Should I reboot repeatedly?
No. Use repeated detection checks to diagnose the unstable hardware path instead.
Can RAID5 run with two disks?
No. Current ZimaOS RAID5 requires at least three drives.
Should I rebuild clean if all disks sometimes disappear?
No. Fix disk detection first; otherwise the new array can fail the same way.
