SCSI 80 pin 146GB Fujitsu Hard Drives locked ID or...

[schmitt330]

Hi,
I have 6 SCSI MAP3147NC 146GB Fujitsu Hard Drives that are having trouble in my dell poweredge 2850.

The situation: I put one hard drive into bay 0 but when I go to configure the raid it registers in bay 3. This is not a problem if I had only one drive but if I put another drive in bay 1 then it also registers in bay 3 then showing no drives in the raid config, like they are cancelling each other out.
All the drives are doing the same thing except in different bays, like putting another drive in bay 4 it registers in bay 0.

To narrow down the problem I have moved the drives around individually to see how they show up and they all seem to register into their own designated bays. By that I mean of my 6 drives, two of them are always seen in bay 4, one is always in bay 3 and three are always shown in bay 0.

My thoughts are that the drives have been programmed in some way to register like this. One obvious way to do this would be to adjust the jumpers, however this drive has none so that is not what was changed.

I have ruled out the server by trying the drives on other servers and the problem is replicated bringing my suspicions more and more to the drives. There is one odd thing I find about the drives though, they are not marked in any specific way to identify what there physical address is, so I am not sure however the previous owner would have organized them. I mean there are no physical difference in the drives except serials.

I have run every format known to googles first 10 pages plus all those found on the ultimate boot cd.

I am dying to know what the problem is. I could easily replace the drives for quite cheap on eBay but I would love to learn what's going on here.
So any help would be great.

Please let me know if you need any information or if I am missing anything.
Also if you are in GA and would like to see the problem in person let me know.

Thanks in advance for any help!!

 

[fzabkar]

I confess that I don't know much about SCSI, but it would appear that your PowerEdge server chassis expects to be able to define the SCSI ID of each device on the backplane by way of 4 SCSI ID pins (pins 39, 40, 79, 80).

Wikipedia states that "the SCSI ID of a device in a drive enclosure that has a backplane is set either by jumpers or by the slot in the enclosure the device is installed into, depending on the model of the enclosure. In the latter case, each slot on the enclosure's back plane delivers control signals to the drive to select a unique SCSI ID."

http://en.wikipedia.org/wiki/SCSI

The following document confirms that the SCSI ID is set by either grounding an ID pin (ID = 1) or leaving it open (ID = 0).

"Configure the SCSI ID by providing the proper open or ground signal inputs to the referenced pins of the drive’s 80-pin new version SCA-2 connector."

Quantum Atlas 10K II Ultra 160/m SCSI Hard Disk Drives:
http://www.seagate.com/staticfiles/maxtor/en_us/documentation/manuals/atlas_10k_ii_product_manual.pdf

These appear to be the relevant Fujitsu manuals:

MAP3147NC/NP / MAP3735NC/NP / MAP3367NC/NP / DISK DRIVES PRODUCT/MAINTENANCE MANUAL:
http://www.andovercg.com/datasheets/fujitsu-map-10k-rpm_prod-manual.pdf

MAW3073, MAW3147, MAW3300 NP/NC SERIES / MAX3036, MAX3073, MAX3147 NP/NC SERIES DISK DRIVES SCSI LOGICAL INTERFACE SPECIFICATIONS:
http://sup.xenya.si/sup/info/fujitsu/diski/c141-c010-01en.pdf

Appendix B of the Product Manual lists the pin definitions for the SCA-2 connector. As in the Atlas case, there are 4 SCSI ID pins (39, 40, 79, 80).

That said, it does appear that your drives are ignoring the backplane pins and are instead identifying themselves with some preprogrammed SCSI ID, possibly stored in flash memory on the PCB. I don't know if it is possible to swap PCBs, or even if it is safe to do so, but that would be one way of positively confirming the location of the ID.

I consulted Fujitsu's SCSI specification, but I couldn't recognise any command that would allow you to reprogram the SCSI ID. Maybe you can see something. FWIW, there is a Set Device Identity command and a Report Device Identity command, but they appear to involve a proprietary scheme of some sort, so I don't know what they do.

Just to add to the confusion, here is an Intel document (plus the erratum):

Intel® Server Chassis SR2400 Technical Product Specification:
http://download.intel.com/support/motherboards/server/chassis/sr2400/sb/sr2400_tps_v20.pdf

Intel® Server Board SE7520JR2, Server Chassis SR1400, Server Chassis SR1450, Server Chassis SR2400 Technical Specification Update:
http://download.intel.com/support/motherboards/server/sb/se7520jr2_sr2400_sr1400_sr1450_technical_spec_upda.pdf

Page 54 of the first PDF shows the 80-pin SCA-2 SCSI Interface Pinout of the backplane connectors. The SCSI ID pins (39, 40, 79, 80) are listed as "Unused".

This begs the question, is there a different version of the SCA-2 standard that provides an alternative way for a drive to identify itself in Intel's servers? Would an Atlas drive be recognised on such a backplane? Would all Atlas drives appear with a SCSI ID of 0 (all 4 ID pins open)?

 

[schmitt330]

Thanks for the detailed write up.

I have yet to test swapping around the pcb boards, that is something that I will try tonight. In theory they should register the same given that I 100% zeroed out the drives, leaving any information on the pcb.

Your conclusion on the Atlas drives not working on the server though is incorrect. I have used Atlas drives before, I have actually used drives with the same part number in the same kind of server. This is why I am adamant that there is something weird going on here.

I will update as soon as I am able to switch around the pcb's.
I will also see if I can perhaps get my hands on another drive with the same part number that is not having this issue and swap pcb's there as well.


Something that I did not include on the original post but I think is perhaps worth mentioning is that the drive is labeled as a 147GB but registers as a 137GB. I have seen drives before do this and thought nothing of it, however these drives are not working so perhaps that has something to do with the problem as well. As from what I know if you completely format a drive, the full labeled size should be shown.


Thanks!

 

[fzabkar]

I was questioning whether the Atlas drives would work on Intel's server, not the PowerEdge. That's because the SCSI ID pins on Intel's servers are "Unused".

As for the capacity anomaly ...

147GB = 137GiB

http://www.google.com/search?q=147+x+10%5E9+bytes+in+gigabytes

See Q1:
http://www.users.on.net/~fzabkar/HDD/HDD_Capacity_FAQ.html

 

[schmitt330]

Best answer selected by schmitt330.

 

[eventhorizons]

Best answer selected by schmitt330.

n
nI am confused.
n
nThe answer selected by schmitt330 contains possible issues and several suggested courses of action.
n
nBut what was "the fix"?
n
nI am having the same issue with some MAP3147NC drives in my Poweredge 1850. I bought 4 drives from an eBay vendor and each one of them shows up as ID #1, regardless of what slot I put them in!
n
nI am enough of an electronic engineer and programmer to know what COULD be happening, but I do not have enough specific experience with these products to know what IS happening.
n
nAnybody have any suggestions?
n
nThanks, Ken

 

[fzabkar]

Ken, are you able to actually measure the voltages on each SCSI ID pin while the drives are installed in the backplane?

 

[eventhorizons]

Actually, now that you suggest that, I'll bet I can!

The connector pins are through-hole and appear on the top of the backplane, right where I can get to them.

Some additional Information:
This is an in-service machine, hosting several of mine and my friends web sites. It is a "new-to-me" Dell Poweredge 1850 that is replacing a 750 that started shutting down randomly.

When it arrived, the backplane was damaged, rendering the second slot unusable. While waiting for the vendor to send me a replacement backplane, I brought the system up with one of the two 37GB drives that it came with. At the same time I ordered the four 147GB drives.

Last night I took it down with the purpose of replacing the backplane and installing the 147GB drives. That was when I discovered the issue.

I considered the possibility that this was either a drive issue or a backplane issue. The 37GB drive appears correctly in whichever slot I plug it into, but the 147GB drives always appear as slot 1. This leads me to think it is a drive issue.

Swapping the PCB's would not help me.

I just bought two more 300GB drives from another vendor on eBay to see about isolating this issue. But I obviously won't see them for several days.

Tonight I will set up and verify the voltage on the pins under a number of test scenarios and post the results. Before that, I will ring out the defective backplane to see if those pins are hard-strapped.

 

[Wolf382]

I have the same issue as Schmitt330. I have 7 Fujitsu 147gb drives trying to install 6 in a PE2850. 3 of then have unique ID#s 0,2,4 but the other 4 all show with ID 3 and when I put them the ID 3s in the empty bays they cancel each other out and don't show.

Has anyone found a solution to clear the ID#s out of the drives.
thanks

Update-Found out that the drives have a HP firmware. not a Dell.

 

[schmitt330]

I have the same issue as Schmitt330. I have 7 Fujitsu 147gb drives trying to install 6 in a PE2850. 3 of then have unique ID#s 0,2,4 but the other 4 all show with ID 3 and when I put them the ID 3s in the empty bays they cancel each other out and don't show.

Has anyone found a solution to clear the ID#s out of the drives.
thanks

Update-Found out that the drives have a HP firmware. not a Dell.

I gave up on the fix. I could buy a newer server for the amount of time I was wasting on figuring it out. Sucks that I couldn't fill my need to answer the mystery but not much time to play around.

 

[eventhorizons]

Update-Found out that the drives have a HP firmware. not a Dell.

I gave up on the fix. I could buy a newer server for the amount of time I was wasting on figuring it out. Sucks that I couldn't fill my need to answer the mystery but not much time to play around.

Same basic result here. I bought some 300GB drives that worked just fine. This is a production machine so I can't waste a lot of time. Would be nice to fix the drives though.