Norco cases and multiple power inputs on backplane

[DoeBoye]

Hello all,

 

<long winded preamble>

I've had my Norco for a while now and have been quite happy with it. Recently (The last month  or two) I have started having random drives red ball on me. At first I assumed it was bad drives/cables/beta UnRaid versions/etc, but smart tests show drives to be perfectly healthy, and after re-introducing the red balled drives to the server and rebuilding, they are fine... And then another drive will randomly red ball... And repeat .

 

So, my question is with regard to Norco cases (mine is the 4224) and the multiple molex power inputs on the backplane. I know that the second input is for a redundant supply, and I know that Norco recommends only using one of them, but I have also read on this forum how some people recommend plugging the PSU into both inputs...

 

Until now I have not used both inputs, as per Norco, but I *have* introduced 2 more drives to my server (note: new drives, as well as old drives have all red balled. New drives are fine). Perhaps there is a limit to how many drives can function with only one molex plug (I'm up to 17 drives -including cache).

</ long winded preamble>

 

So, actual question: Is the random red balling of healthy drives behaviour that others have seen when using only one molex plug on the Norco backplane?.

 

Thanks!

 

[EDIT]: And if not, what were people seeing before they switched to using both power inputs!

 

 

[Chris Pollard]

What is your power supply?

[wreck]

What PSU are you using and what kind of drives do you have (5400? 7200? green?)? 17 drives is a lot, but not unreasonable for the right PSU.

[marcusone]

1 molex per plane (4 drives I believe) which is fine for 1 molex.

Likely the red balling is due to a faulty back plane. Which norco seeks to have reoccurring issues with.

 

In theory if all 4 drives were 7200rpm and spun up at the same time you MIGHT exceed the recommended amp rating of the molex. But I've never had that issue and have 4 7200rpm drives on a single molex in several systems.

 

Sent from my SGH-I727R using Tapatalk 2

[DoeBoye]

Hey All!,

 

The power supply is a single 62 amp rail 750w Corsair. It is fairly new as I swapped with a buddy less then a year ago because his new PSU didn't play nice with his mobo, but my older one had no issues with it...

 

The drives are: 13x green WD EARS/EADS and 5400 RPM Hitachis, 3x 7200 RPM drives (2 data and Parity), and 1 SSD cache drive.

 

I considered the backplanes to be the culprit when this first started happening, but as the drives that have been failing have been spread over multiple backplanes, my gut tells me that I should look at the simpler possibilities first (Good ol Occam!). It's definitely possible that multiple backplanes are failing simultaneously, but I figured I'd isolate the easy things first.

 

I know there are some folks (Couldn't find the thread, but I'm thinking Raj might be one of them... Raj? ) that are big proponents of the multiple plugs per backplane setup, so that's why I wanted to find out if anyone who has switched to that did so because of similar behaviour...

 

Not to sidetrack my original question, but if it does turn out to be the backplanes that are faulty, what's Norco's warranty policy. The case is coming up to two years old now, so pretty sure it's out of warranty...

 

I'm also testing another theory that came to me from threads of people with similar issues, but I won't bring it up yet in order to stay focused on my original question

 

So again, to summarize, who switched/uses multiple plugs per backplane, and did you see any odd behaviour before you went to it that was solved by doubling up...

 

Thanks guys!

 

[EDIT] sp/gr

 

[EDIT 2] de- . I always use too many of those buggers.

 

[EDIT 3] PSU Specs and link, specific drives

[ohlwiler]

I posted some info in the following two threads that may help:

 

http://lime-technology.com/forum/index.php?topic=15773.0

 

http://lime-technology.com/forum/index.php?topic=16180.0

[DoeBoye]

I posted some info in the following two threads that may help:

 

http://lime-technology.com/forum/index.php?topic=15773.0

 

http://lime-technology.com/forum/index.php?topic=16180.0

 

Thanks! That was v. helpful! Sounds like I might be on the right track . I've messaged jmos1277 for an update to his issue. Let's hope it was resolved by the extra plugs...  *crossing fingers*.

 

Anyone else have any experience with solving drive issues by plugging in both molex connectors to every backplane?

[lionelhutz]

Do you have extra cables you can plug in to test with? You can read and speculate all you want, but the proof would be in testing it.

[DoeBoye]

Do you have extra cables you can plug in to test with? You can read and speculate all you want, but the proof would be in testing it.

 

You're absolutely right! Unfortunately, I am short 3 plugs, but I have ordered more, so should be getting them in early next week.

 

To be honest, I am hoping to get some more instances of this... phenomenon (?)... together in one thread so that if it does in fact fix the issue, others in the future will be able to find it easily. If it doesn't fix it, then I guess we're no further ahead then before...

 

The theory does make sense to me though, so I'm looking forward to see what happens when I get the new cables. The 'good' news is that I can pretty much get it to fail every time by doing a parity check, so once the new cables are in, it should be easy to judge whether the problem is fixed or not.

[DoeBoye]

Another user pm'd me asking if this was resolved, so I thought I'd post an update if anyone else has a similar issue:

 

So the sata-to-ide power cable switchers came in, and I hooked up a cable to both sets of molex plugs.

 

That seemed to reduce my dropouts substantially. Since I doubled things up, I have only had one occurrence of a falsely red-balled drive.

 

Unfortunately, I can't say for sure if that was solely the issue, as it still did it once since I made the change. It did reduce the event SUBSTANTIALLY though.

 

Perhaps it may simply be that every now and then, too much is going on (something getting par-checked/unrared, while a large transfer is happening), and a controller has a minor hiccup.

 

I recently replaced my supermicro mv8 cards with m1015s, and I'm curious to see if I ever see another one of these random red-balls. Maybe the lower available bandwidth of the supermicro card (4x vs 8x pci-e) was to blame, even though on paper it should not matter...

 

Unfortunately, it is almost impossible to diagnose such a rare issue. It could even be a combination of the particular parts I have in my server not always playing well together...

 

In conclusion, for anyone having this same issue, I did see a definite improvement from doubling up the power connections to the backplanes...

 

HTH!

[JohnnyBra]

Thanks DoeBoye!

 

I've been having exactly the same symptoms out of my Norco case, and was suspecting either Backpanes or Controller Cards.  Your thread was the first I started suspecting power.  Looking inside, I had one of those 1x7 molex adapters, and was essentially running everything off of one PSU lead.  And now that I'm looking, it was the Backpanes with the most non-green drives attached that we're the biggest offenders.

 

Spreading out my heavy drives to separate backpanes, and have ordered some sata to molex adapters to help bulk up my power input leads.  Happening once in a blue moon, I can handle. 

 

Lately it had been wiping out every time I run Parity Check, and I've since just powered down until I had time to look into it.

[DoeBoye]

Sounds like it may be the same issue! It's certainly a fairly inexpensive way to determine if it is.

 

Definitely try and get as many separate PS cables as you can powering the backplanes. I don't know if it is making a difference, but I tried to make sure each pair of molex plugs was powered by separate cables from the PS.

 

Good luck!

 

Don't forget (like I did  ) to post back your results to this thread! Based on my findings, and the findings of others before me, this may definitely be of help to those with similar issues.

 

[RickInHouston]

Had one drive not even being able to be pre-cleared in top right slot of my Norco 4224.

 

Precleared a new disk in another slot, moved it to the upper right slot and no blue light on the drive cage but it is seen in my GUI.  Don't want to test fate so I moved the drive to another bay and the blue light is intense and being recognized as a good drive.

 

Does this mean the blade of the Norco is bad?  It is the fourth WD Red 3TB drive on that blade.  Power supply is 650Watts.  Only have five drives currently in the Norco.

 

I think I will just bypass that bay and never use it.  If I do contact Norco, is replacing the blade a difficult task?  I guess everything must be disconnected t get to it...