SuperMicro 16 bay SAS Backplane only recognizing 12 bays

[doccie]

Hi,

I recently purchased two refurbished Supermicro SC836TQ (3U, 16bay) chassis with SAS backplanes (Standard SAS836-TQ planes).
One works perfectly fine, but the other fails to recognize/initialize any disks installed in right-most column.

Unfortunately, I'm a little out of my depth in this, so I'm looking for common causes and/or ways to troubleshoot or fix.

Any help would be greatly appreciated!

 

[i386]

- make sure that all the molex connectors are correctly plugged in
- check the backplane with known working cables*
(- use high quality cables (I had problems in the past with cheap >75cm cables))

*Edit: Connect a sideband connector from the backplane to your hba and query the enclosure information if there are problems

 

[doccie]

Thanks for the reply!

Switched the fourth row cables (SATA + sideband) with the third row, but same results.
But for the life of me, I wasn't able to pull out the molex... So next, I'm going to try and get a voltage readout from the molex connector to see if it still works.

Any tips/commands for getting the enclosure information?

 

[doccie]

Tested the molex using some copper wire and a voltage meter... the molex was functioning correctly.
Any other tips you might have?

 

[Allan74]

Perhaps a little 'entry-level' in terms of Info, as I am no Pro, but, are you using all SAS ? All SATA ? or a Mix ?

I have personal experience with something similar with a couple single sff-8087 connected 12xdrive 6Gbps Intel backplanes that support SAS/SATA simultaneously, but require the SAS drives ALL to be populated consecutively starting from a specific end of the backplane if using mixed drives.

In my case, I can begin populating SATA drives starting from Bay#1, 2, 3, 4 etc....but populating SAS drives, I am required to start at bay#12, then #11, 10, 9 etc....and the same population rules apply even if I am using the same drive interface, but leaving empty bays.

 

[doccie]

Thanks for the insight! Unfortunately using all SATA drives if I'm not mistaken. Will double-check once I get home though.
I started filling up drives from #1 to #13 and noticed that #13 was not functioning. I tried inserting in #14 and #16 as well, but with the same result.

 

[Dave Corder]

Just to clarify, do the disks that don't work in the right column also not work in any other bays, or do they work elsewhere?

Also, what do you mean by "Switched the fourth row cables (SATA + sideband) with the third row, but same results."? What was the same? That the last column didn't work?

I have an 836 TQ backplane in one of my JBOD units - no issues with a mix of SAS and SATA drives, connected to a Dell SAS2 expander.

 

[doccie]

The disks work fine in another column or my other SC836.
I tried to switch the SFF > 4xSATA cables from the third column with the cables from the fourth column, to rule out problems with the cable.
The result of that experiment was that the third column continued to work, while the fourth column continued to not work, so ruling out that it is a problem with the cables.

And just checked; all SATA drives indeed.

 

[itronin]

my recollection from looking at the tq backplane (after removal) and some of the traces is that I felt each molex powered the column of drives that were above the connector. I think you've confirmed that. Each column looked more or less like a cookie cutter version of the previous (typical design pattern kind of stuff I said to myself). I also see there are some large caps by each drive. Do you see any leakage or discoloration with the caps in the misbehaving column?

I will just leave this next bit right here.

I have a spare BPN-SAS-836TQ backplane and will have a spare BPN-SAS2-836EL1 backplane sometime in the next 60 days.
The TQ backplane was working when I pulled it 4 years ago. The expander backplane is in use at the moment but I'll be switching that system back to a TQ backplane sometime soonish. There's also one reasonably priced TQ on the bay for about 90USD shipped.

 

[itronin]

Tested the molex using some copper wire and a voltage meter... the molex was functioning correctly.
Any other tips you might have?

oh one other thing and worth a shot. try swapping the molex in the misbehaving column with a molex from another (working) column. I see you tested the voltage but sometimes under load they can behave differently. If the misbehaving column works then you've ruled out the backplane and are likely looking at the chassis pdu which is replaceable.

 

[sth2100]

When you put the drives in the 4th column, do the drives spin up at all? If they spin up, but not detected, then you at least know the power is reaching that side. If the drives don't spin up at all, then you know it's something likely with that molex cable, or the caps as was previously mentioned.