... I think I am going to go back. Mostly just because I can return these CoolerMasters but would have to go through the trouble of selling the Thermaltakes on eBay.

 

I understand completely.  In addition to the simplicity of returning vs. "e-bay'ing", it IS more convenient to have the hot-swaps.  I haven't replaced them on my older systems -- I just don't use them anymore on newer systems because I love the airflow of the CoolerMasters.

 

I'll definitely get drive temp numbers though

 

Definitely post the comparison !! 

By the way, the best of both worlds WAS available in a system I built for a friend ~ 18 months ago.    We used IcyDock 5-in-3 cages to mount 15 drives (could have done 20, but we just used 3 cages as that's all we needed) in a Lian-Li PC-P80 case.

 

That's an amazing case -- it has 3 140mm fans mounted on a swinging front door.  The hot-swaps mount in the 5.25" bays, but when you close the door these fans are blowing air directly into them 

 

Unfortunately, that case is no longer available.   

http://www.newegg.com/Product/Product.aspx?Item=N82E16811112159

 

For up to 12 drives, there's an excellent alternative (I've been thinking of using this case for another friend who's toying with UnRAID) in the PC-A76.  It has 12 internal drive bays, with 3 140mm fans blowing directly over them.  They're not hot-swap, but changing a drive is almost as simple => you just pop off the cables; flip a locking tab; then pull the drives out.  You screw 4 screws on a drive, then just push it in place.

http://www.newegg.com/Product/Product.aspx?Item=N82E16811112388

Well I know you all have been waiting with baited breath on what the status is of my particular issue... I believe this is solved!! w0ot. Power supply replacement FTW.

 

I'm gonna keep an eye on it for a while before I authoritatively state that it's solved but it's been running pretty solid since I swapped out the PSU last Wed (and replaced the crap HDD). I'm back to the Thermaltake enclosures and things seem good.

 

Gary, I'll get you temps tomorrow but they're going to be pretty inaccurate because my ambient has been changing quite a bit as of late. Yaay spring.

 

THANK YOU ALL SO MUCH FOR YOUR HELP! Seriously, fantastic community, really appreciate it.

-Dan

Excellent news ==>  I'm sure you're VERY relieved 

 

... glad your hot-swaps weren't, in fact, the problem.    I DO like hot-swap cages ... I just like the temps better in cases or cages that have direct, large airflow on the drives.  If I was going to build a new UnRAID server now, I'd use the Lian-Li PC-A76 for up to 15 drives (12 in the built-in bays plus a 3-in-2 for the 2 5.25" bays);  or the Lian-Li PC Q25B (I already have one of these) for up to 7 drives [this is an IDEAL case for an UnRAID Plus system  ]

 

With 4TB available now, and 5TB WD Reds coming by the end of the year, those two choices should be PLENTY for just about anyone !!  [30TB or 70TB capacity !!]

 

Great to hear. I had the same problems but after a psu upgrade I have been running without any issues for 10 days now. And believe me I have been torturing the server trying to make it fail

Great to hear. I had the same problems but after a psu upgrade I have been running without any issues for 10 days now. And believe me I have been torturing the server trying to make it fail

 

It's amazing how many problems that SEEM like failing/failed disks are actually either cables or power 

Well to be fair the drives are actually failing, but only because of insufficient power from the sub par psu. 

Well to be fair the drives are actually failing, but only because of insufficient power from the sub par psu.

 

Semantics 

I wouldn't say a drive that's no plugged in, or is plugged in but has insufficient power,  has "failed" ... it's certainly "not working" ... but for a good reason.

 

Just wanted to give everyone that's helped me out a quick update. I did switch back to the Thermaltakes and immediately got a couple HDD errors. So I swapped BACK to the CoolerMasters and been rock solid ever since. For those of you that were counting that was three issues that needed resolving: 1) Bad PSU, 2) Bad drive enclosures, 3) Bad HDD. But we got there! Thanks for all your help.

 

FYI, here are the temps I'm seeing with the CoolerMasters:

 

 

Thanks y'all! I'm super duper happy!

Nice temps  :)

 

[but as I'm sure you know, I'm not surprised  ]

I did switch back to the Thermaltakes and immediately got a couple HDD errors. So I swapped BACK to the CoolerMasters and been rock solid ever since.

 

When I first installed the iStarUSA BPN-350 cages I was getting disk errors.  These went away after I updated the firmware on my USAS2 controller card.

Ugh. Not solved. Still having intermittent HDD problems. Parity disk, data disk, doesn't matter. Seem to drop off at random. Re-seated cables. Doesn't seem to matter. This is getting REALLY ridiculous really quickly. Do I have to rebuild from scratch? Seems insane. I'll get a new syslog up here once it happens again... which will be soon I'm sure.

Thanks y'all! I'm super duper happy!

 

Ugh. Not solved.  ...  This is getting REALLY ridiculous really quickly.

 

Well, that was a mighty short period of "super duper happiness" !! 

 

And clearly a VERY frustrating issue.

 

Just to recap ...

 

=>  You've replaced all the SATA cables with new locking cables.

=>  You took additional connectors out of the mix by replacing the hot-swap cages with CoolerMaster enclosures

=>  You upgraded to a high-quality Seasonic X-series PSU

=>  You replaced a bad drive

 

Is that all correct?

 

At one point, you indicated you were going to add a quality UPS w/AVR to the system.  Did you do this yet?    If not, that's the next thing I'd do -- a UPS SHOULD be part of your server anyway, so I'd do that before making other changes.    Even if you have reliable power from your utility (with few outages), you almost certainly are still subject to brownouts and over-volts as power demand fluctuates ... a UPS keeps those from being "seen" by your computer.

 

Any good UPS with AVR will do ... the one I suggested earlier is a good choice [http://www.newegg.com/Product/Product.aspx?Item=N82E16842101381 ]

 

Just in case a UPS still doesn't resolve this, I reviewed you system configuration info -- is this list still accurate?

 

Proc - Intel Celeron G530 2.4GHz LGA 1155

Mobo - GIGABYTE GA-H61M-HD2 Micro ATX

Memory - 4GB Patriot DDR3

SATA Interface Cards: 2 x SYBA SI-PEX40064 PCI-Express 1x SATA III Controller (Marvell 88SE9215 Chipset) with 4 ports each.

 

Processor is fine. 

Motherboard is pretty low-end, but should still be fine. 

Memory is okay as long as you're using standard 1.5v modules -- is the 4GB a single module; or a pair of 2GB modules?

The SATA interface cards are okay; but connecting 4 drives to a PCIe x1 slot is a bottleneck -- shouldn't cause any issues except slower speeds

 

One thing you can do [perhaps while waiting for a UPS to be delivered  ]  is to download the latest MemTest+ [http://www.memtest.org/ ] and run it for an extended period [24 hrs would be good]    It doesn't seem very likely, but you could be having a very intermittent memory failure.

 

If I was going to replace anything in your configuration, it would be the SATA cards.  I'd put in an 8-port x4 or x8 controller.  But that should NOT be necessary !!

 

Just to recap ...

 

=>  You've replaced all the SATA cables with new locking cables.

=>  You took additional connectors out of the mix by replacing the hot-swap cages with CoolerMaster enclosures

=>  You upgraded to a high-quality Seasonic X-series PSU

=>  You replaced a bad drive

 

Is that all correct?

 

That is all correct, yes. I'll try the mem test as you mentioned above and report back. Right now I'm running the array without the parity drive as it seems when I am doing a parity-sync is when the issues show up. Maybe that points to memory. I don't know but I'll try anything at this point.

Originally you had problems with drives attached to any controller, including the mobo controller - is that still the case?

 

I was never sure why I had problems when I added my drive cages - as far as I am aware, there is no additional electronics involved (* see note below) - the cage is simply an extension of the SATA cable.  It was only drives on my USAS2 controller which were affected.

 

I concluded that the problem was caused by a small timing difference, or a slight loading on the data SATA signals - I was ecstatic when the very next firmware update for the controller card made the problem go away.

 

* Note:

Actually, I realise that there must be some additional electronics for the cages to implement the drive activity lights.

Originally you had problems with drives attached to any controller, including the mobo controller - is that still the case?

 

Yes... in the last 24 hours I have had issues with drives on all three controllers (mobo and one on each of the pci controllers). I hope it's a memory issue. That should be fairly straightforward to fix.

 

I'll also look into BIOS and firmware updates for the motherboard and cards respectively.

as far as I am aware, there is no additional electronics involved (* see note below) - the cage is simply an extension of the SATA cable.

 

* Note: Actually, I realise that there must be some additional electronics for the cages to implement the drive activity lights.

 

In addition to the circuitry to implement the activity lights, using hot-swap cages also introduces additional connectors.  ANY connector causes some electrical loss, so if there are marginal signals or a bit too much "noise" in the connection it can have adverse effects.

 

 

Yes... in the last 24 hours I have had issues with drives on all three controllers (mobo and one on each of the pci controllers). I hope it's a memory issue. That should be fairly straightforward to fix.

 

I'll also look into BIOS and firmware updates for the motherboard and cards respectively.

 

The fact you're seeing problems with drives on ALL of the controllers is, in a way, a "good" thing -- it tends to exonerate the controllers as an issue.    I don't think it's the case; but there IS a test you could do if you want to take the time to do so:  using PCIe x1 controllers with 4 drives attached is a definite bottleneck.  I don't think the timing implications of that are what's causing your issue;  but at this point who knows !!    You can at least eliminate that as an issue by reconfiguring your array with only 2 drives on each of these controllers (just build a new config with fewer drives).    Obviously that would take a few days to test ... reset the config (unplug the unused drives from the controllers just to be sure they weren't in the picture at all);  compute parity; then do a parity check; etc.

 

 

... by the way, is there a UPS on the system now?

 

Hi All:

Just came back to see what was shakin and excited for the release of RC15. Going to give it a go this week. I also wanted to provide some more data for all of you awesome troubleshooters to see if you can help me continue to pinpoint my continuing problem.

 

I've been running unprotected, without a parity drive, since May 21st. Zero problems whatsoever. Runs great. As soon as I add that parity drive in, problems galore. Same stuff I've seen previously. Random drives drop out. I'm wondering what you all think this could be pointing to. Seems like it could be that my parity drive itself is having issues. Would this cause other drives to drop out?

 

Gary, I've not run the memory test yet and I haven't put it on a UPS. I know I know. I need to do these things. Especially if you think this behavior points to a memory issue. I'll try and run it sometime this week. Just not looking forward to the 24hrs+ of downtime.

 

Thanks y'all!

Another possibility is that the extra drive is just enough to over-tax your power supply.  Something like that would cause drives to randomly drop out.

Another possibility is that the extra drive is just enough to over-tax your power supply.  Something like that would cause drives to randomly drop out.

 

That's what I would think too... which is why I upgraded to the Seasonic 850 Gold PSU (http://www.newegg.com/Product/Product.aspx?Item=N82E16817151102) and I would think that would be enough to power my 12 drives...? I hope! I don't think my wife is going to let me buy another PSU.

You are right - that power supply should power more than 12 drives.  Perhaps it is worth checking the cabling again - especially if you have splitters in place.

You are right - that power supply should power more than 12 drives.  Perhaps it is worth checking the cabling again - especially if you have splitters in place.

 

Worth a shot. Server hasn't been touched though and I would think if it were a cabling problem it would manifest itself with the parity drive disconnected as well. Super strange that it's rock solid sans parity drive.

Super strange that it's rock solid sans parity drive.

 

Not necessarily all that strange.  When you're computing and/or checking parity is probably the ONLY time you're actually using all 12 drives at once.  [A drive rebuild also does this; but clearly you aren't doing any rebuilds without good parity]