Migrating slowly...

[sdumas]

OK - I screwed up royally and I have decided to upgrade to Beta 5 14 / start from fresh.

 

I won't go into the tribulations of my demise - but I wanted to know if the following is possible and how.

 

I have 10 drives in a previous unRAID 4.7 installation (no longer working properly) that I would like to use on a Beta 5 environment.

 

Now - the way I want to do this is to build a parallel system with Beta 5 14 and use two brand new 2 TB drives to start and get me up and running.

 

Once my new Beta 5 14 is up and running, I would like to take SOME of the drives (with data on them - no user shares) and populate the Beta 5 system.

 

First - is this possible?

Second - if it is, by putting the old 4.7 drive in the 5.0 array, will it be recongized? Will I have to do an initconfig? Will I be able to keep my data?

Third - I will have a brand new parity drive - everytime I add an "old" new drive, will I have to run parity or I can wait until a few drives are in. If I need to run parity everytime, It will take me almost two weeks to get the array back up and running.

 

Any suggestions, clever ideas, anything?

 

Thanks!

[cpshoemake]

Yes, it is possible. You can add multiple drives and use initconfig to reset the configuration and then rebuild parity. By doing this however, you are implicitly trusting the validity of the data on and the health of these drives. What do you mean by "not working properly"?

[sdumas]

"not working properly"...  You asked for it - here is my saga

 

I was at 4.5.4 and got a drive with a red ball (drive 5).

 

I went along and do as I usually do and got a replacement drive for it (from 320g to 2 TB). So far, so good.

 

I reboot the server and then I get an other drive with a red ball... yikes. Obviously now the system does not want to start and I get the too many missing drive error.

 

I get a little panicky and perform a reiserfsck on the drives. They both gave me errors... oops two drives gone... not good.

 

This is where I started to do stupid things...

 

I upgraded to 4.7.

 

I started to switch drives around thinking it could be wiring or backplanes issues (Norco 4220 on Asus mobo with SuperMicro SATA 8 port card)

 

The drives showed the same behaviour no matter where they are. Now I think that the drives are really bad.

 

I decided to put the original 320 gig in the box and then it gave me the "drive too small" error.

 

Damn if I do and damn if I don't - I can't rebuild since I have two bad drives and I can't go back since it does not want to accept it.

 

At this point - I already gave up and decided that the two drives are gone and data irreplacably gone too. (BTW - I do not use user shares - thank god)

 

Decided to format the drives... after a few manipulations, booted the server and drives showed up as "unformatted" (two of them).

 

I click on the box - yes I want to really do this - and click on format.

 

Well - thankfully - nothing happened - the drives did not format.

 

I get really frustrated and replace both drives with new ones try to format - no go.

 

Now I can't get the array to come back in ANY state at all.

 

I put the original drives back in the machine and perform an initconfig.

 

The drives come back all blue - hurray - I got it - or so I thought.

 

No longer do I press the start button - I am starting to get errors all over the place - the parity drive get orange ball.

 

Now I am pretty sure I have hardware problems other than my drives. Mostly the Supermicro controller is a prime suspect.

 

Changed the supermicro controller with my older controllers  (2x PCie Sil3132? + 1 PCI Promise TX300) - now the damn machine does not want to boot anymore. BIOS errors with not enough PCI interrupts something (remembered this issue with ASUS from a previous post) arrrgghhhh.

 

I get really pissed off and went to the store to buy a brand new motherboard (Gigabyte GA-Z68XP-UD3) hoping that the interrupt issue will be gone.

 

Install the new motherboard - reinstall the older controllers - and connect 4 of my existing drives from the old onboard mobo controller to the new one.

 

Boot the machine (remember this is still 4.7 installed) and I am missing 4 drives. All the ones connected to the onboard controller. Have to come to the conclusion that 4.7 does NOT support the onbard controller of the Gigabyte mobo.

 

At this point - it's late - I have spent over 36 hours on this damn thing and I just want to get it over with. (I am thinking making a video of me thowing the server and the drives into a nearby river - running it over with my car - digging a hole and pouring cement over it and plenty of other ideas really...) I get one last spark of an idea.

 

I decided to remove all drives from the machine and install the two new 2 TB in a brand new 5.0 Beta 14 configuration.

 

I reboot the machine with a clean config - 2 drives on the onboard controller of the gigabyte mobo - and.... nothing. The drives are still not recognized... arrghhhh - it's now 2 in the morning... getting tired and stupid... (more stupid if it can be possible)

 

I upgrade the BIOS of the mobo; went from F2 to F8.

 

Reboot... The machine stops at a PCI discovery and hangs there... does not want to boot anymore... tears in my eyes - fire in my soul... The river idea starts to become quite plausible...

 

I decided to fidget with the BIOS of the mobo. Disable USB3, make sure the controllers were in IDE mode and a few more tweaks. Desperation here.

 

I rebooted and ... what a surprise the drives show up - unRAID 5 Beta 14 boot and it shows my drives ... (can you hear the angels sound in the background...).

 

Now I have nothing to freakin' loose. What the hell I am one hair away from going insane - why not cross the line...

 

I remove one of the two new drives (remember 2 brand new 2TB drives) and decided to install in the case my two original "bad drives" from my previous installation.

 

The system boots and guess what? - the two drives showed up unscathed in the array. I do the dance of joy!

 

I now have a brand new parity drive - and my two original supposedly bad drives back on a new array running 5.0 Beta 14 and parity is being rebuilt.

 

I will be adding the rest of my drives today...

 

"I got a feeeeling, that today is gonna be a good day, that today is gonna be a good good day... I got a feeling - wooo hooo...."

 

[cpshoemake]

Wow, what a nightmare!

 

If it were me, I would add all the drives that had not redballed, rebuild parity, copy the data off the two drives that unraid previously redballed, and then run a few iterations of preclear on them to make sure they are good to go before adding them back to the array.

 

The way I understand it is that unraid redballs a drive when it has a write failure. So, while everything may appear fine for awhile, you could be in the same boat again later if an area of the drive is unwriteable.

[sdumas]

Three and a half days later - it's solved - kind of. (just a few more steps)

 

I have lost a drive worth of data (1 TB). Couldn't recup it in any way and now it is not recognized by any tools... (reiserfsck sees no partition - rebuild-tree sees nothing - dd sees nothing... ) I have to consider it dead. oh well...

 

Gotta love Norco...

 

Loose backplanes made data iffy on drives. I was getting different results on almost each reboot - something to hit your head on the wall for.

 

Started to tighten screws on the backplanes - no differences - loosen screws - reset the boards - removed a board (broke it of course... actually not broken but does not want to fit in anymore - the little pressure plate that help hold it into place decided to fold over and is stuck at the back of the black bracket making the board not going completely to the end ... useless now really...).

 

Was still getting weird results. But one drive was consistently "screwing" up - DRDY error (drive not ready) - removed it and the errors kind of went away.

 

Now was the time to recreate parity - used a Seagate 2TB that I pre-cleared a couple of days earlier - but - since I changed config and did a bunch of initconfig so often along the way - it was seen as a bad disk... again - invalid disk -  (I thought about getting it a handicapped licence plate in hope it could help but... LOL).

 

I had an extra motherboard - assembled a quick PC, installed the basic version of unRAID, took three extra drives I had and I am doing a preclear on all of them.

 

Removed the parity drive from the array for now - until one of the three drives finished preclearing. The array is now not protected but up and running with almost all its data. With the bad parity drive - the parity was being rebuilt, but at a rate of 1 Mb/s - estimated time to finish - 88650 minutes... That's a little bit over two months... Did not think it was very practical... Call me impatient but I did not want to wait that long.

 

Still a little ways to go and I am hoping I can see the light at the end of the tunnel.

 

It all started because I wanted to take a 320gb drive to a 2TB drive... then all hell broke loose.

 

As the old saying goes - "When you're up to your neck in alligators - it's difficult to remember that the initial goal was to drain the swamp..."

 

After parity goes in and is fine - I will change that drive again... hopefully this time it won't create havoc.