New drive died

I just installed a 4 TB HGST DeskstarNAS (ALE640 version) drive that I purchased it almost a year ago but it's been sitting on my desk. I put it in the server and ran 2 rounds of pre-clear on it (via gfjardim's preclear plugin). I didn't examine the results in detail, but there were no reallocated sectors. After that finished, I added it to my array, and have been using it for about 2 weeks (since 9/16, to be exact).

 

My wife woke me up about 3 this morning asking if I smelled something funny. It was a burning plastic smell, that I finally traced to the server. I took a look at the Main page and saw that my new drive was listed as "Not Installed". I immediately shut it down (via the WebGUI, not the power button), and the smell began to dissipate.

 

When I got home this evening, I powered the server back up, and it seems to be running OK, except that the drive is still listed as "Not Installed". At least I'm not getting the burning smell.

 

I'm attaching the preclear report for the drive, but as you'll see, it's looking pretty useless. I'm not certain what happened there. I'm also attaching the syslog which I just grabbed. I realize that it's only since this evening's boot, but at 3:30am, I wasn't thinking clearly enough to grab the log before I shut down the server. I don't know if this will help diagnose the issue or not.

 

It appears that the server is currently running by reproducing the data from the missing drive by spinning all the rest, including the parity drive, so I'm good for a little while. I've got a new disk on order, and will check to see if the dead one is still under warranty. It should be as it has a 3-year warranty.

 

• Since the server doesn't think the drive is there, I may as well physically remove it, correct?
  
• Does anyone see anything in the syslog that might indicate what the issue was?
  
• When the new drive gets here, I'll run a couple of preclears on it again. Any idea what happened to the preclear report?
  
• Once the preclear is done, all I need to do is stop the array and assign the new drive to the old drive's slot and it should commence rebuilding, correct?
  
• Anything else anyone recommend I check on?
  

syslog_2016.10.06.txt

preclear_report_PL1331LAHE2R0H_2016.09.17_21.15.37.txt

Something wrong with the preclear report, nothing useful there.

 

Since you only have 1 parity drive you are currently running without parity protection. If another drive fails both of the failed drives data is lost.

Yup, that's why the replacement is on its way. FedEx says it will be here by tonight. Unfortunately, 2 preclear cycles will be about 100 hours, so I've got that fighting me.

 

For some reason, HGST's web site says my SN is invalid for the old drive. I'll have to physically pull it tonight & read it off the label, as well as looking at the electronic one that was reported by unRAID (and captured in a screenshot I've got of my drive setup).

I'm no expert, but I'm pretty sure this is a Bad Thing^TM

 

 

It's the result of this

 

 

The smell of that melting is what woke my wife up and probably prevented the house from burning down. Yay wife!!!

 

---

I figured out the SN issue - unRAID reported 6 extra characters from the electronic SN that aren't printed on the label on the drive. Once I took those extra letters off, HGST's web site recognized the SN and tells me that I've got another 18+ months of warranty. I'm wondering, if I wipe that soot off, will they send me a replacement?

My guess is you got that SATA power plugged in backwards at the other end. Is it a MOLEX plug?

It is a Molex to SATA adapter. Molex connectors are keyed, it would take a whole lot of force to plug it in backwards! I'd think that it would have gone nearly immediately if that were the case though - 5v -> 12v and 12v -> 5v. Especially the 12->5v side of things wouldn't have gone well.

 

This has been running for better than 2 weeks, including a 2-pass preclear.

It is a Molex to SATA adapter. Molex connectors are keyed, it would take a whole lot of force to plug it in backwards! I'd think that it would have gone nearly immediately if that were the case though - 5v -> 12v and 12v -> 5v. Especially the 12->5v side of things wouldn't have gone well.

 

This has been running for better than 2 weeks, including a 2-pass preclear.

Well if it was working then probably not backwards. There have been people on this forum who immediately fried a drive because they were able to force the molex to plug backwards.

It is a Molex to SATA adapter. Molex connectors are keyed, it would take a whole lot of force to plug it in backwards! I'd think that it would have gone nearly immediately if that were the case though - 5v -> 12v and 12v -> 5v. Especially the 12->5v side of things wouldn't have gone well.

 

This has been running for better than 2 weeks, including a 2-pass preclear.

Well if it was working then probably not backwards. There have been people on this forum who immediately fried a drive because they were able to force the molex to plug backwards.

 

So,

A) I've passed trurl's intelligence test (knows MOLEX plugs are keyed) 

B) Holy Cow! People force those things on backwards?? Never even thought to try it!

Never use Molex to SATA power adapters, most of them are poorly manufactured and can be extremely dangerous. Get a better power supply with more SATA connectors or a modular one where you can plug in more proper SATA leads.

 

This guy does a deep analysis into the one that caught fire in his server.

https://www.youtube.com/watch?v=fAyy_WOSdVc

 

Original fire video

https://www.youtube.com/watch?v=EYkof-csPfI

 

EDIT: not sure why my links aren't showing? had to put them in code tags

I know what I'm doing when I get home! 

Or, buy real high tolerance good spec name brand stuff. It's really not expensive, you just have to shop at the right places. Granted, this post is USA centric.

http://www.molex.com/molex/products/datasheet.jsp?part=active/0887511310_CABLE_ASSEMBLIES.xml

http://www.mouser.com/ProductDetail/Molex/88751-1310/?qs=sGAEpiMZZMvAZU35sbsmiVyE4XRYrcARfdS4sKNhMUs%3d

I watched the YouTube video and it's clear that the failure is at the SATA end of the cable. The Molex end is not the problem at all.

I watched the YouTube video and it's clear that the failure is at the SATA end of the cable. The Molex end is not the problem at all.

 

Not really the bit with the cables touching each other are kinda due to the 4 thick wires coming from the molex end, a normal sata connector doesn't have such thick wires, but its more pointing at the manufacturing quality of these cables not purely the molex variety. IMO there's no reason to be using molex connectors these days unless you are using older hardware and absolutely cannot avoid it. I am unsure whether a pure sata splitter (1 male sata to 2 female sata) would be included in this but would you want to take the risk? I wouldn't.

Never use Molex to SATA power adapters, most of them are poorly manufactured and can be extremely dangerous. Get a better power supply with more SATA connectors or a modular one where you can plug in more proper SATA leads.

 

This guy does a deep analysis into the one that caught fire in his server.

https://www.youtube.com/watch?v=fAyy_WOSdVc

 

Original fire video

https://www.youtube.com/watch?v=EYkof-csPfI

 

EDIT: not sure why my links aren't showing? had to put them in code tags

 

Lovely advice, so let me ask you. Where do I get extra SATA power cables for my Corsair TX650M?

 

Sure, it's not the fanciest PSU on the planet, but it was recommended here on the forums when I bought it and it's a modular design with a nice little plug in for the SATA power. Unfortunately, the only thing I can find via Google shopping is reconditioned PSUs, 8-pin to PCI-e cable sets, or complete sets of custom built cables for $140+ (more than I paid for the PSU in the first place).

 

It's got more than enough oomph for the drives I've got plugged in and plenty of room to grow.

 

At least

Or, buy real high tolerance good spec name brand stuff. It's really not expensive, you just have to shop at the right places. Granted, this post is USA centric.

http://www.molex.com/molex/products/datasheet.jsp?part=active/0887511310_CABLE_ASSEMBLIES.xml

http://www.mouser.com/ProductDetail/Molex/88751-1310/?qs=sGAEpiMZZMvAZU35sbsmiVyE4XRYrcARfdS4sKNhMUs%3d

provided useful advice with an actual recommendation of a quality product.

 

(and US-centric works for me, at least...)

I watched the YouTube video and it's clear that the failure is at the SATA end of the cable. The Molex end is not the problem at all.

 

Not really the bit with the cables touching each other are kinda due to the 4 thick wires coming from the molex end, a normal sata connector doesn't have such thick wires, but its more pointing at the manufacturing quality of these cables not purely the molex variety. IMO there's no reason to be using molex connectors these days unless you are using older hardware and absolutely cannot avoid it. I am unsure whether a pure sata splitter (1 male sata to 2 female sata) would be included in this but would you want to take the risk? I wouldn't.

 

I completely disagree with you. Thick power wires are a good thing as they have a lower resistance. The failure in the video was entirely due to bad design/construction at the SATA end. The fact that the other end of the cable had a Molex connector is totally unconnected with the failure. Also, the fact that it's a splitter is totally unconnected with the failure.

 

Never use Molex to SATA power adapters, most of them are poorly manufactured and can be extremely dangerous. Get a better power supply with more SATA connectors or a modular one where you can plug in more proper SATA leads.

 

Mea culpa...

 

I was looking for something totally unrelated, when Amazon popped this up as a suggested item: https://www.amazon.com/gp/product/B00HBGN0F8/ref=oh_aui_detailpage_o00_s00?ie=UTF8&psc=1 Turns out it's the Corsair CP-8920113 which has 4 SATA power connectors and, after some searching Corsair's site, finally giving up, creating a support ticket and waiting a day, will work just fine in my TX650M PSU. That said, a suggestion of where to find such a device would have been appreciated.

 

Additionally, I cleaned the power contacts on the drive I pulled out of my server, looked at them under a magnifying glass, and cleaned them some more (with rubbing alcohol & a pair of tweezers to get the little junk out). I plugged the drive into a SATA/USB3.0 adapter, plugged it into my Win10 machine and presto, it recognized the drive immediately!

 

Of course, Win doesn't recognize the format, but the HGST test tool finds it. I ran the short SMART report, and it looks OK. I'm running a long SMART test on it currently. If that comes back OK, I'll plug it in to my unRAID via the adapter and do a couple of preclear passes. If it makes it through that OK, I think it'll be safe to add back into the array.

The drive passed both the short & long SMART tests.

 

I've plugged it into the server via a USB dock, and Unassigned Devices recognized it. It gave me the option to run a file system check, which I've done. The results are:

FS: xfs

/sbin/xfs_repair -n /dev/sdp1 2>&1

Phase 1 - find and verify superblock...
Phase 2 - using internal log
- zero log...
- scan filesystem freespace and inode maps...
sb_fdblocks 712741510, counted 712749702
- found root inode chunk
Phase 3 - for each AG...
- scan (but don't clear) agi unlinked lists...
- process known inodes and perform inode discovery...
- agno = 0
- agno = 1
- agno = 2
- agno = 3
- process newly discovered inodes...
Phase 4 - check for duplicate blocks...
- setting up duplicate extent list...
- check for inodes claiming duplicate blocks...
- agno = 0
- agno = 1
- agno = 2
- agno = 3
No modify flag set, skipping phase 5
Phase 6 - check inode connectivity...
- traversing filesystem ...
- traversal finished ...
- moving disconnected inodes to lost+found ...
Phase 7 - verify link counts...
No modify flag set, skipping filesystem flush and exiting.

The penultimate line in Phase 2 looks a bit questionable to me:

sb_fdblocks 712741510, counted 712749702

A) Would someone more versed in file system checking take a look at this, please?

 

While unRAID does seem to recognize the drive and that it's formatted with xfs, UD doesn't seem to want to mount it. I click the "mount" button (both the one next to the drive itself, and the one next to the mount point), and, after refreshing the web page, the button is still labeled "mount", and attempting to access the mount point from my Win machine yields nothing.

 

My plan, as noted above, is to simply run a couple of pre-clear passes to ensure that it still seems to be in good shape. I don't think that there's any advantage of attempting to get the drive mounted via UD, since the data that's on it seems to have been successfully rebuilt via unRAID when I put a replacement drive in.

B) Does this seem to be a reasonable approach?

I would have recommended something fully modular, like something from Seasonic:

 

http://www.tweaktown.com/reviews/5800/seasonic-m12ii-750w-evo-edition-750-watt-80-plus-bronze-psu-review/index3.html

 

 

For instance, this 750W model, while it only includes cabling for 10 SATA drives on 3 connectors, can potentially be extended with extra cables to power a total of 16 SATA drives, four on each port.

 

A larger PSU is probably warranted for much more than 16 drives. Maybe. They seem to assume when you're buying a PSU this large, that you intend to 3x SLI/CrossFire, not hoard data.

 

Still, I looked at the photos and specifications for your PSU, and it looks like it's very minimally modular. I'd have to see a label for the four connectors I saw to determine if they're capable of hosting SATA power.

 

These Seasonic products go as far as even making the base ATX 24+4 and the 4+4 CPU power cables detached. Maybe a little overkill, but still neat.

Well, the guy from Corsair says it'll work, so I'll take him at his word. If the whole thing melts down, then I've got their email to say they told me it would work. Of course, I don't have everything backed up, but the critical stuff is. (I can live without a couple of TV shows...)