Sun Oracle F40/80 PCIe Flash Accelerator - Mute Offline Uncorrectable SMART Error SPAM in Unraid

Preface

The following write up pertains to my F80 card setup as a RAID10 CACHE drive in Unraid 6.8.3. Other version of Unraid may have different settings. Disabling any SMART notifications should only be done on the Sun Oracle flash cards, and not on other hard drives.

Introduction

The Sun Oracle Flash cards are great little solid state drives. They have many uses, but are particularly useful in Unraid servers. I have some setup as cache drives, running dockers.

When I first added an F80 to my Unraid server, everything worked great, but I slowly started to get SMART errors. “Oh crap, I got a bad card”, was the first thing that came to my mind. Luckily the fine people on the SB Discord calmed me down, said it is a “known issue” with these Cards and the SMART errors could be safely ignored. But what could not be ignored was the constant SPAM from UNRAID.

Not wanting to completely turn off SMART, I researched the notification and found I could selectively disable the notification for the incorrectly reported SMART value. With notifications off, all is well, drive reports normal, and Unraid stopped spamming me.

What the SPAM Reporting Means

Disk health SMART value 198 is for “Uncorrectable sector count”, reporting a problem with reading and/or writing to a sector. This is important on mechanical drives as it may imply damage or wearing of the disk surface.

The "Known Issue"

The F40/F80 cards to do not report SMART Attribute 198 correctly. I’ve never found the definitive answer as to why, but some speculate it is related to how attribute 198 is being interpreted from the controller on the card. The same speculators have shown their cards to be 100% healthy with other tools (https://docs.oracle.com/cd/E41278_01/html/E41251/z400000e1790781.html) and no problems present in production. This error is safe to ignore.

How the SPAM presents in Unraid

When your F40/F80 misreports SMART errors, you will have two, obvious symptoms. The first is a thumbs down for drive health:

The second is a constant, nagging SPAM about Uncorrectable errors:

image

Telling Unraid to Shut Up (Muting the Error Notification).

Since we know it is okay to ignore this error, muting is pretty simple.

  1. Log into Unraid web interface and go to MAIN tab.
  2. Identify each of the drives on your F40/F80. Remember, each card has four SSD modules.
  3. Click the hyperlink to open device details (in my example CACHE 2)
    image
  4. Scroll to section “SMART Settings” and uncheck the box for “Attribute = 198 Uncorrectable sector count”
    image
  5. Hit the APPLY button, then done.
  6. Repeat Steps 3 to 5 for each of the remaining devices on your F40/F80 card.
  7. Make a sandwich, you did it.

Conclusion

Your Unraid is now SPAM Free (at least from the F40/F80 cards). Always monitor health of ALL your drives regularly. Oh, and backups, make backups.

5 Likes