We've got a new SATA system in recently, and I seem to be having a fun time trying to identify which drives are what on the devices panel.

 

I setup the drives one by one earlier today, finding out 2 of my 8 new drives were DOA... Once I had determined which drives were good, I inserted one drive at a time, from start to bottom, and booted up the unraid server with each added disk. This allowed me to pick the top slot as parity, second slot as disk 1, etc.

 

Well, after I did all this (but didn't format or bring the array up in any way), I transported the system to it's new home, only to find that when I booted it back up, that no disks were listed in the main settings page. So, I go to the devices page, and all of the device ID's are showing up as "not found" (or something similar) after the device "location". When I chose from the dropdown menu, I have six "new" drives to chose from with different "locations".

 

So... I start the process over, powering down and installing only top drive... assign as parity, power down, etc.

 

Well, I get thru the first 4 drives and then drive 5 won't show up. So I power down, wiggle the drive bay, no luck. So I open the case, wiggle the power cords and SATA cables... And boot up to no drives found on the main page. 

 

I am now on my THIRD time thru setting up the drives (currently on disk 5) and am concerned that this may happen again AFTER I've had the system up and running with data on it. This is a royal pain.

 

My major concern is that none of the device ID's seem to make much sense... THIS time thru I have:

/dev/scsi/sdh9-9c0i0l0

/dev/scsi/sdh10-ac0i0l0

/dev/scsi/sdh9-9c0i1l0

/dev/scsi/sdh10-ac0i1l0

/dev/scsi/sdh4-4c0i0l0

/dev/scsi/sdh2-2c0i0l0

(I typed these in by hand, I think I got that all right!)

 

BUT, before, when the drives weren't appearing in the main settings page, I am 90% sure (should have written that down) that the parity drive was coming up as "sdh5" and that other drives were coming up "shd6".

 

And for the record, I haven't traded slots with any drives... they get pulled an inch out of their chasis, but not removed, so they're not getting stuck in different bays.

 

Can anyone here give me an idea of what might be going on?

 

Thanks

jakev,

Please make sure the PCI controller cards are well seated.

 

Your technique for identifying the drives is ok.  A faster way would be to look at the sticker on each disk and jot down the disk serial numbers - you probably won't have to write down the entire s/n, just say the last 3 or 4 characters/digits.  The s/n will be displayed on the Devices page along with the disk identifier.

 

I agree the disk ID's seem to not make much sense.  We're working on a section of the website that describes how to build an unRAID server which includes a section on SATA device naming in linux.  Hopefully that will explain the method behind the madness.

Thanks for the idea on serial numbers.

 

I hope this is something that can't happen once the array is defined and a parity check has happened... as it has only happened before this point. (knock on wood)

 

The next time I power down, I will check to make sure the cards are seated correctly... but I'm not pressing my luck till I HAVE to restart.

 

After that last time of wiggling the cables inernally, everything seemed to go fine, except I found out one more drive was essentially dead... during the parity check I stopped it after around 50 or 60K errors... changed slots and tried again, with the same results. But I'm glad to see the errors right away, rather than after data is being shuffled around. Whew.

 

 

Well... I seem to be back at it again...

 

After a week of the system being down (PSU problem?)... which I'm still working out the solution, ugh.

 

I went to startup the system after the outtage, and now my devices have changed... again.

 

Before I go and make any changes or try to set things up, I wanted to list my old "devices" page settings, and then I'll start to post the new settings. I have no idea why this changes back and forth.

 

Actually, I noticed that my previous post has the settings I was using... so now, with only the parity drive installed, I have as my only option: /dev/scsi/sdh5-5c0i0l0 (sdb1) SATA:ST3320620AS/(ser#)

 

So, essentially, everything stayed the same, except the "9's" turned into "5"s"... pretty much making everything screwed up.

 

I have yet to figure out why it seems to change... or how to make it stop changing.

 

Does anyone have any idea  what the problem is, and how I can get it to stop?

And as usual... actually asking someone else/posting on forum/attempting to contact support leads to (a possible) solution.

 

I managed to get the devices back to how they were on my original post... by plugging in the second power supply. I can't quite figure out how having both PSU's plugged in would change the device ID's... unless somehow the cages for the removable HD trays make the system change how it behaves (even without any drives in them).

 

I actually have about 3 hours of a parity sync that needs done, and then I can THINK ABOUT trying to confirm this issue/problem by restarting with only one PSU. As of now, this is my theory... FWIW.

 

Also, I think I may have stumbled upon a solution to another possible "issue" with the prebuilt unraid servers... These have no power switch, as that is removed to make room for the 12 HD trays. Soooo... in order to turn on the system, you have to unplug both PSU's then plug them back in, so that the system thinks you "lost" power... then it will start up when it "regains" power.

 

Well, that's all fine and dandy until you do like I (think I) did, several weeks ago and tweaked the settings in the BIOS. I am fairly certain I managed to change the "on power loss" setting in the BIOS to "last state". Which I did because hey... I was going to be out of town for a while and I didn't want the system to power itself on if it REALLY lost power... Well, my oh so smooth self powered down the system (see where this is going?) to leave for a few days... and when I get back, unplugging/plugging in the PSU's caused them to start up for half a second, then go dead in a blaze of glory. Ahhhh... After cutting up the pretty wireties holding all the molex cables in place... I found the new PSU does (almost) the same thing. So I decided to drag an empty case over next to the unraid case, and use the power switch cable to connect it to the mobo, and viola, we have a system running again.

 

This is also a theory, as I'm not going to go back thru the hassle of putting the original PSU back in place, plus I actually like having a real power switch, so I won't be doing the old fashioned "unplug/plugin" dance to power on the unraid server. If anyone else would like to test this theory out... make sure you have a power switch you can plug into the mobo before you try it! Don't say I didn't warn you.

 

Man... what a hassle. hopefully SOMEBODY can learn from my mistake(s).