Drives keep dying (Bad luck, or am I killing them?)

[thatflashcat]

I was hoping I wouldn't need to return here for a while, but alas, I've had yet another drive fail on me. At first I assumed it was just another bad drive before noticing a similarity to the last failure (which was just a month ago): Both drives were connected to the last power sata power connector (top-most in the picture below).

Is it bad practice to use three spinning disks on a single connector like this? I'm using an EVGA 750w G3, so I didn't have another way of connecting the drives and never thought it could be an issue.

All told out of 6 drives:

• 2 drives have shown smart errors
• 3 drives had died completely

This leads me to my next point: My smart tests have only included 2 drives out of 6 for the past year or so... For whatever reason, the smart task only included selected drives rather than all of them. I'm guessing the menu changed since I set this up 2 years ago.

Regardless, one of the three dead drives that has died completely was a replacement for a previous failure. This makes it less likely that the failing drives are lemons from the same batch, but I guess I could be unlucky twice. Is it possible that my PSU is killing them?

(Lastly, temperatures are in the 30-35 C range, so I doubt that is the issue.)

 

[Jailer]

I think I'd be replacing that power supply with something else.

 

[thatflashcat]

I think I'd be replacing that power supply with something else.

Is it the most likely culprit? I don't want to replace the PSU for a *maybe* fix. Or could it be something to do with the power connectors?

Also,
I have a second pool in the same box (same PSU) that has had zero issues.

 

[thatflashcat]

Update:
I pulled the drive and put it in my external drive reader (just tested with a working drive) and I hear a lovely chirp/click as the drive spins up.

 

[sretalla]

Is it the most likely culprit? I don't want to replace the PSU for a *maybe* fix. Or could it be something to do with the power connectors?

Could be.

Maybe it's just one of the rails that's no good in the PSU... consider testing it or replace.

 

[Arwen]

@thatflashcat - It would help if you listed the drive models, (and the forum rules suggest the whole system, including software version).

 

[thatflashcat]

@thatflashcat - It would help if you listed the drive models, (and the forum rules suggest the whole system, including software version).

Sorry, I didn't see this until now! I'm using six Seagate ES.3 ST4000NM0033 drives. I'm thinking the way they were packaged (minimal padding) may be causing problems now.

Freenas specs:

CPU: Intel Xeon E5-2650 V2
Cooler: Some noctua cooler
Motherboard: Supermicro X9 SRL-F
RAM: 8 * 8GB Samsung DDR3 ECC 1333
Controller: LSI H220
Case: Phanteks Enthoo Pro
PSU: EVGA 650w G3

Maybe it's just one of the rails that's no good in the PSU... consider testing it or replace.

Test it how, exactly?

 

[Apollo]

I don't think cables would cause the disk to fail unless the cable is in such a bad shape it is causing the disk to constantly reset itself, but then again this would be seen from the system as the drive is being disconnected.
Harddrives, like any electronic system are taking power and shape it to meet their need, there are switching regulators and/or LDO to provide power to the individual circuits on the drive. Most likely the drive has what is called a voltage surpervisor constantly monitoring the voltage levels. In the event of a brownout or under voltage/ overcurrent condition, the voltage supervisor would reset the MCU/DSP on the drive.
Where this could be problematic is if the drive is resetting when the heads aren't parked and the heads could land on the platters. But then again, the drive would enter a reset state and the SATA controller would/should loose connection until the drive comes back online after coming out of reset.

So, unless the drive constantly disconenct from the system, the cable shouldn't be the cause of the issue.

 

[sretalla]

Test it how, exactly?

Using a multi-meter and one of the molex connectors? You'll need to check for a wiring diagram to see which rail applies to which connector(s) and see if you can get the same reading from each of the rails.

 

[joeschmuck]

You have a modular power supply based on EVGA 650w G3. If the only drive failures are related to the last connector on the power cable then the odds are the cable was built poorly and one of the cables were not crimped properly, it could be in the last connector or the second to last connector. The point is that you will unlikely be able to repair that specific cable.

I see that you have two options if this is actually the case:
1) Contact the manufacturer and tell them that you have a defective SATA power cable. Odds are they will ship you one for no cost.
2) Use the PERIF cable which provides a few 4-pin MOLEX connectors and buy a 4-Pin Molex to SATA adapter (I like the Molex to two SATA pigtail myself) and this will allow you to test your diagnosis and also give you the option to run the system as-is while waiting on a new cable form option 1.

Option 3: If you are handy and can solder, you might be able to repair the SATA power connector but only if you know exactly what you are doing.

If Option 1 or 2 does not fix the problem then power might not be your issue but all I have to trust that you have a good conclusion for now.

If you had all drives failing on that one power cable then I'd say the problem is with the PSU, but that is not what you have indicated.

Is it bad practice to use three spinning disks on a single connector like this?

No, that is perfectly fine to connect three drives to this connector, it was designed for it.

 

[thatflashcat]

Using a multi-meter and one of the molex connectors? You'll need to check for a wiring diagram to see which rail applies to which connector(s) and see if you can get the same reading from each of the rails.

Oh, that. :P

Contact the manufacturer and tell them that you have a defective SATA power cable. Odds are they will ship you one for no cost.

I have plenty of spares so I can swap the drive cables, but it seems odd that both cables would have issues.

I'm thinking the drives were damaged in transit.

Speaking of drive cables, I'm out out of SATA and PERIF connectors (my six drives are on SATA, my boot and a second pool are on PERIF). What's the best option for connecting more drives? (I have plans to turn my freenas box into a proxmox vm and pass the drive controller to it, so I expect to add drives).

Sorry for the late replies, the website outage hit when I was trying to reply. :P

 

[kherr]

I had 3 drives fail in 6 months, 2 of them within a week. I opened the case .. unplugged the sata cables ... put the LSI card in a new slot .. then plugged everything back in. Then I wiped the partitions off one of the drives and used it to replace a 'failed drive '. After resilvering I ran a srcub,
no errors, SMART info is being read (before SMART would not be read), that was 2 months ago.

 

[thatflashcat]

I had 3 drives fail in 6 months, 2 of them within a week. I opened the case .. unplugged the sata cables ... put the LSI card in a new slot .. then plugged everything back in. Then I wiped the partitions off one of the drives and used it to replace a 'failed drive '. After resilvering I ran a srcub,
no errors, SMART info is being read (before SMART would not be read), that was 2 months ago.

So one of the drives actually wasn't dead?

 

[joeschmuck]

I'm thinking the drives were damaged in transit.

Using the data you provided which was two drives both failed when connected to the exact same power connector but worked fine in another system leads me to believe the problem is with the power cable. I don't see how it could be anything else given that very specific data.

I have plenty of spares so I can swap the drive cables

You do? You must have purchased several of the same power supplies then, or how did you get these power cables?

Speaking of drive cables, I'm out out of SATA and PERIF connectors (my six drives are on SATA, my boot and a second pool are on PERIF). What's the best option for connecting more drives?

Buy a power supply with the proper number of power connectors. you could get away with a few 'Y' harnesses but using too many could put a strain on your power cable(s), the connectors, or power supply.

 

[thatflashcat]

Using the data you provided which was two drives both failed when connected to the exact same power connector but worked fine in another system leads me to believe the problem is with the power cable. I don't see how it could be anything else given that very specific data.

No, two drives on two different (but identical) power connectors. The drives happened to be attached in the same position on both connectors.

You do? You must have purchased several of the same power supplies then, or how did you get these power cables?

These are common EVGA power cables. I have plenty of extras from the 750W G3 in my desktop and a 750W G2 in another system.

Buy a power supply with the proper number of power connectors. you could get away with a few 'Y' harnesses but using too many could put a strain on your power cable(s), the connectors, or power supply.

Maybe, but power supplies with more power connectors also have much higher wattages than I need. I'm already using a 650W PSU.

 

[kherr]

So one of the drives actually wasn't dead?

It appears NONE of them are .... faulty wiring connections seems to have caused the errors. Once cleared of being tagged as 'faulty' TN excepts them. Still no R/W/C errors and the SMART data is being read when before there were notifications that it wasn't readable.

 

[thatflashcat]

It appears NONE of them are .... faulty wiring connections seems to have caused the errors. Once cleared of being tagged as 'faulty' TN excepts them. Still no R/W/C errors and the SMART data is being read when before there were notifications that it wasn't readable.

Odd. Unfortunately, mine really are dead and won't show up on any system.

 

[joeschmuck]

Unfortunately, mine really are dead and won't show up on any system.

So your drives are bad, it has nothing to do with the power connection location.

You can see how contradicting information can confuse us (your quote below)

Update:
I pulled the drive and put it in my external drive reader (just tested with a working drive) and I hear a lovely chirp/click as the drive spins up.

If you need anymore help on this topic, please be clear in what the current problem is, what you have tested, and did I say, be clear.

Cheers,
-Joe

 

[thatflashcat]

So your drives are bad, it has nothing to do with the power connection location.

You can see how contradicting information can confuse us (your quote below)


If you need anymore help on this topic, please be clear in what the current problem is, what you have tested, and did I say, be clear.

Cheers,
-Joe

Sorry for the confusion!

The cable question was about if they could have caused the drives to fail. At this point, I have a replacement for the latest failure on the way. Hopefully it's the last for a while!