8TB Drive drops out of JBOD volume on 6.4.0 on Ultra4

[powellandy1]

Hi

I have 2x8TB Seagate Archive HDDs in JBOD array in a Ultra4 on 6.4.0 production.

I started doing a rsync backup (this device was the destination) and after about 10min the second drive showed up as failed.

The device still reported 'healthy' on the first page, the volume size was still 14TB and the backup continued OK.

I took the drive out and ran SeaTools - it passed the self test, short test and long test with no problems.

Logs emailed.

I know these drives, when subjected to a sustained write, can slow down periodically - is the NAS mis-interpreting this as a failure??

Thanks

A

[StephenB]

If the backup succeeded, then the drive clearly wasn't dead.  Probably someone from Netgear would need to troubleshoot the false status though.

If you have a 14 TiB volume, then you aren't running JBOD.  You must be spanning both drives with RAID-0.  The risk there is that a failure of either drive will lose all the data on both.  Two 8 TB volumes are much safer.

[powellandy1]

It says JBOD on the status page. Each drive is 7.2 formatted I think so 14.4 in total.

Edit:

From volume.log

Pool backup:
	Device: /dev/md126
	Node: 9:127
	HostID: 5e26b038 (native)
	UUID: 635f4fc7-2874-40a5-965f-1d738131aaf7
	Mount point: /backup
	Size: 15616260544KB (14892 GB)
	Available: 15615978408KB (14892 GB)
	RAID Level: JBOD
	State: unprotected
	Action: idle
	Flags 0x148
	Type: btrfs
	RAIDs:
		md127
			Size: 15618353664
			Level: 0
			Action: idle
			Members: 1
		md126
			Size: 15618353664
			Level: 1
			Action: idle
			Members: 1

 It was also quick to set up and add second disk - few minutes - whereas I think (although I may be wrong) to set up RAID would take a few hours to initialise an array - especially with these disks.

[StephenB]

Do you have two volumes or only one?

My guess is only one from the log.

If one, it is spanning RAID-0 no matter what volume.log says.  And if either disk fails you lose the data on both.

[powellandy1]

Only 1 volume.

So a test would be to pull one and see what happens??

Thanks

A

[StephenB]

---

@powellandy1 wrote:

So a test would be to pull one and see what happens??

---

I wouldn't recommend that unless you want to lose your data.

[Skywalker]

---

@StephenB wrote:

Do you have two volumes or only one?

My guess is only one from the log.

If one, it is spanning RAID-0 no matter what volume.log says.  And if either disk fails you lose the data on both.

---

Not exactly true.  It's not striping, so it's not RAID 0; but it is concatenated into the same volume "pool".  Flex-RAID allows you to do this.

[powellandy1]

Thanks for the clarification. Whilst I know they are technically not supported they do seem to be causing a heap of problems - https://community.netgear.com/t5/Using-your-ReadyNAS/Seagate-8TB-Archive-drives-ST8000AS0002-do-work... - that seem to be worse with 6.4.0, which might be related to the new linux kernel.

A

[Skywalker]

Right, and it certainly doesn't appear to be specific to ReadyNAS.  There are many posts about the same issue online, with an official kernel bug report still open at https://bugzilla.kernel.org/show_bug.cgi?id=93581.

[powellandy1]

Thanks.

Could you tell me the best way to get my ultra4 to 6.2.4 (currently on 6.4.0) - do I have to go back to 4.2.27 and then on to 6.2.4 or can I go back to 6.2.4 via GUI and factory reset (or do I need to USB recovery)??

Or shall I wait for a 6.4.1 (is this something Netgear will look at further??)
A

[powellandy1]

Back on 6.2.4 and seems to be OK. Doing a 9TB backup so we'll see...

A

[Skywalker]

We'll have something for 6.4.1 to try to address this.  However, we don't have any SMR hard drives to test with, so we won't be able to verify the fix.

[powellandy1]

Thank you. If i'm feeling brave i'll try it 🙂
A

[powellandy1]

I have another 8TB drive now and in the next couple of days an RN102 should arrive. I am happy to test something if it would be helpful.

Quick related question - If one drive in a JBOD array dies do I lose the whole array or (if a new drive is replaced) just what was actually on the bad drive?? My understanding of JBOD was it should be just that which is on the bad drive is lost but other answers suggest the whole array might be lost...

Thanks

Andy

[StephenB]

---

@powellandy1 wrote:

Quick related question - If one drive in a JBOD array dies do I lose the whole array or (if a new drive is replaced) just what was actually on the bad drive?? M

---

There is no such thing as a JBOD array in a ReadyNAS.

If you have one volume for every disk, then you have JBOD - all the disks are independent, so if you lose one you only lose the data on that disk.  There is no array, just a bunch of disks.

If you have one volume spanning all disks, then you have a RAID-0 array, and not JBOD.  In that case if you lose any drive, all the data on the volume is lost.

[powellandy1]

Just for followup.

Finally bit the bullet and put 6.4.2T59 on my Ultra6 with 3x8TB SMR drives in.

Just written 200GB over 40min and it's fine. No errors. No stops. No increase in command timeouts.

It looks like the fix that was added in 6.4.1 works.

Thanks all.

Andy