Multiple drive purchased with the same zero serial number wont work in Truenas

[ziggygt]

I purchased 3x 4TB white label drives from an eBay vendor to use in my Truenas server. I have a weird issue that they all report zeros for the serial number. They show the zeros in Crystal Disk Info program (Win 10) and when installed in Truenas. Each drives tests OK when installed separately and they have less than 30 hours of runtime so they appear "new" as advertised. When I format and install the 3 drives in Windows 10 (for test) they all get a drive letter and can all be accessed in the OS but Crystal Disk Info sees only one. In Truenas only one drive shows up for use in a pool. If I install one drive at a time the drives work OK in Truenas, but if install them together only one shows up. I have never seen this before. I thought all drives were manufactured with a unique serial number? I think Truenas and Crystal disk expect that.

I wonder if the eprom data on these drive modified improperly to show as "new" or was there a missed manufacturing step? The vendor is a reputable one. I have purchased other new, but old date code drives from them before with no issues. I really don't want to return them if there is a way to update the serial numbers or work around the issue.

I hope someone can shed light on this issue. Is there something I am missing? Is it time to get an RMA from the seller?

 

[mr44er]

I have a weird issue that they all report zeros for the serial number.

That's fishy and really bad if this prevents you from creating pools.

I thought all drives were manufactured with a unique serial number?

Yes.

I hope someone can shed light on this issue.

Refurbing drives is one thing, maybe this went wrong or it's the intention and he is hiding that these disks are stolen. Could be also the reseller, I don't know.

Is it time to get an RMA from the seller?

Yes, ASAP!

 

[elvisimprsntr]

Smells fishy. There are OEM and other tools that can reset the drive SMART data, including the serial number.
I would be suspicious the evilBay seller is selling used drives as new, and intentionally did not set the serial number.

Lesson learned: Don't buy drives from sketchy sellers on evilBay or anywhere else.

 

[ziggygt]

I have requested an RMA. This truly a mess because the drives were advertised as 7200RPM 64MB cache White Label, but they are in fact 5980RPM 8MB drives so It should be possible to return as "not what was advertised". Working with the seller. I usually buy only old date code/new Hitachi drives, they seem to last forever. I thought I could live with the 5980RPM but I just noticed the cache size issue as well.

This whole mess has taken my server down but I pulled all the drives on the pools I had precious data on so they are protected from the mayhem.

I was trying to upgrade a 1x4TB+3x2TB Z1 pool and combine it with a 4x4TB Z1 pool to make an 8x4TB Z2 pool.I am glad I moved all the data from the pools to be combined before attempting this. I had planned to upgrade the 2TB drive pool one drive at a time as practice for when my 4X6TB Z2 pool needs expansion. This is probably why some people prefer Unraid over Truenas.

 

[ziggygt]

The supplier took back the white label bad drives. I purchased 4 "new" date code 2013 Hitachi 4TB drives from a different eBay supplier. I checked them out with crystal drive info. No funky serial number or status. They installed perfectly in Truenas. I asked the original supplier for an explanation on how they sent the wrong drives and how messed up they were, that I needed some explanation if I was ever going to buy from them again. All I got back was " In future, please drop me a message once you have placed any new order. I will request our warehouse to check your drive before ship". I did not feel totally satisfied by this.

 

[mr44er]

I did not feel totally satisfied by this.

I'm assuming he's just another reseller and it doesn't have to have been intentional. If he buys a large batch of used hard drives somewhere, he can't possibly test them all individually. The profit margin would be too low for that, unless he specializes in it and does refurbing properly or certified. Of course, it could also be that the hard drives are stolen goods or that the BIOS chip is simply broken and the seller didn't know it himself. In any case, it is important that he took the hard drives back without any problems. I would give him a second chance in the future, but not a third if things go stupid again.

 

[ziggygt]

I'm assuming he's just another reseller and it doesn't have to have been intentional. If he buys a large batch of used hard drives somewhere, he can't possibly test them all individually. The profit margin would be too low for that, unless he specializes in it and does refurbing properly or certified. Of course, it could also be that the hard drives are stolen goods or that the BIOS chip is simply broken and the seller didn't know it himself. In any case, it is important that he took the hard drives back without any problems. I would give him a second chance in the future, but not a third if things go stupid again.

I purchased many drives from that supplier in the past. I did not post the name as I was not trying to trash them in these posts just understand how Truenas responded to identical/blank serial numbers. The drives worked OK if my desktop.

 

[Terry Kennedy]

I'm assuming he's just another reseller and it doesn't have to have been intentional. If he buys a large batch of used hard drives somewhere, he can't possibly test them all individually. The profit margin would be too low for that, unless he specializes in it and does refurbing properly or certified.

This can happen by honest mistake or due to seller dishonesty. There are a few sellers who have been known for years to reset the SMART data and sell worn-out drives as new.

Quite a few years ago I purchased a bunch of Dell 2.5" SAS drives from a huge seller (they currently have over 40,000 active eBay listings) who said they were unused pulls. When I got them, my routine testing showed that they had 30K+ POH and most had grown defects. They were very proactive in dealing with the problem and told me they got them directly from Dell, who represented them as being unused pulls. They had Dell send me an RMA label (so this was really Dell's fault), immediately refunded my money, and sent me a couple of $50 Starbucks gift cards for my trouble.

 

[ziggygt]

This can happen by honest mistake or due to seller dishonesty. There are a few sellers who have been known for years to reset the SMART data and sell worn-out drives as new.

Quite a few years ago I purchased a bunch of Dell 2.5" SAS drives from a huge seller (they currently have over 40,000 active eBay listings) who said they were unused pulls. When I got them, my routine testing showed that they had 30K+ POH and most had grown defects. They were very proactive in dealing with the problem and told me they got them directly from Dell, who represented them as being unused pulls. They had Dell send me an RMA label (so this was really Dell's fault), immediately refunded my money, and sent me a couple of $50 Starbucks gift cards for my trouble.

Thanks for the insight. I did want to trash the supplier because I have good buying experience in the past. Part of the reason I posted was that Truenas was very confused with the duplicate serial numbers as was Crystal Disk. Windows did not have a problem. I thought this was interesting and confusing.

 

[ericloewe]

The TrueNAS middleware requires unique serial numbers to track disks and tends to break in weird and wonderful ways when this is not the case. This is an issue that typically pops up with virtual disks, for instance.
I'm not sure anybody's ever examined exactly how things break, because it's never really been an interesting or useful problem to solve, but the underlying layers (Linux or FreeBSD, ZFS, etc.) don't care about the serial numbers, it's just that the middleware gets really confused and tends to be unusable.

 

[ziggygt]

It seemed to randomly get confused as to what drive was which even in pools that did not contain the weird drives. I wisely had unplugged the drives in pools that contain data I care about so they were not affected. If I had not unplugged them, tears would have been involved. Your comment about virtual drives is an interesting one.

 

[ericloewe]

If I had not unplugged them, tears would have been involved.

Probably not. Big asterisk on that one, don't just go trying that on a production machine, etc. But: Since the middleware doesn't really do much to the pools in normal operation, it should be possible to just disconnect the offending disks and be back to normal.

 

[sam55todd]

It could be possible that internally devices are identified by Hardware_ID within low-lever operating environments and in this specific case having same serial_Ids and other characteristics device-side firmware/bios - this hardware_ID derived by systems becomes the same for all devices - therefore system can't distinguish between drives and consequently reports back incorrectly.

 

[ericloewe]

What are you talking about? USB, ATA, SCSI and NVMe all support a clear, unambiguous serial number. The only potential issue, aside from virtual disks, is with the dodgier end of the USB spectrum - but any disk/adapter worth using will support UASP these days and report whatever the ATA/NVMe serial number is via the SCSI layer.

 

[ziggygt]

What are you talking about? USB, ATA, SCSI and NVMe all support a clear, unambiguous serial number. The only potential issue, aside from virtual disks, is with the dodgier end of the USB spectrum - but any disk/adapter worth using will support UASP these days and report whatever the ATA/NVMe serial number is via the SCSI layer.

The drives were Hitachi 4TB spinners attached to HBA. Drives were returned.

 

[ericloewe]

Sounds like their internal configuration had been wiped, presumably to clear old SMART data in the process. Extremely dodgy.