Hi all,

 

I'm currently running an eTrayz nas in RAID 1 mode,

I don't have enough room left on the device, so I bought some new hardware to build an UNRAID storage server.

I currently have 2x2TB (2TB data) hd's in RAID 1 and want to go to 4x2TB (6TB data) in the new system.

I'm using a separate OS disk, to install a full slackware on using the tutorials in the wiki. However this disk has not yet arrived from delivery.

 

However since I want to use the 2 disk from the old array in the new system, can I just plugin one of the RAID 1 disks in the new system, mount it and transfer the data (the eTrayz has REALY slow transfer speeds) to the other 3 (precleared) disks. And later preclear that disk and add it as parity. I don't care that much if for some unfortunate event a disk crashes before the parity disk is added, since its mostly just movies and series for my HTPC.

 

I found some info on how to mount it, but that requires mdadm to be present. I did the preclearing in my HTPC running from an unraid usb stick. However the stick doesn't have mdadm on it. Can I install that or would it be better if I installed it on the full slackware disk when it arrives.

 

Once all the parts of the new system have arrived, I would like to get it running asap and not wait for an other 26 hours of preclearing just so I can start the transfer.

 

I hope I explained my situation clearly, and you can help/advise me TNX

 

ps. Little background info on myself. I manage several Debian (web)servers, so I know my way around linux (I'm not an expert on it, but i know how to install, configure and maintain it from console), and I've never run Slackware before.

If your etrayz is using linux software raid (hinted at my mdadm in your post) then I don't think you'll be able to do what you want in slackware.

 

The unraid kernel driver effectively 'replaces' the stock kernel md driver and so making you unable to be able to do any non-unraid md operations.

 

That's my understanding anyway.

 

*However* you might be able to have one of your data disks precleared and formatted ready for unraid using the unraid kernel module, then reboot preventing the unraid module from loading - so having the stock linux kernel md module available - mount the newly formatted unraid data disk as a normal reiserfs and mount your raid-1 using mdadm.

 

Then copy like that.

 

Once done, reboot back into the unraid kernel module and start your array - you'd have to rebuild parity.

 

That *might* work, or some slant on it. But you'd have to think about it / be very careful

If your etrayz is using linux software raid (hinted at my mdadm in your post)

Yes it is software raid

 

I did some more searching online.

I can delete the raid metadata from the hd with a `dmraid -rE`. The hd should be a normal hd again.

And I can mount the partitions normally in slackware (right?)

 

Tonight (CEST) I will attempt this, first removing one hd from the array and checking to see if I can mount it.

If that succeeds, than I will preclear the other hd.

---

I want to copy the data to a unraid user share, so the data(folders) will be nicely spread acros the disks.

At least thats my understanding of what unraid does.

 

The hd used to transfer the data will later become the parity disk

My limited understanding of dmraid is that it's for handling hardware / fake software (i.e still on a card) arrays.

 

*not* linux md software raid arrays.

 

I'd be very sure before you run that command - though chances are it will just do nothing if it can't find a signature it knows about on the disks.

My limited understanding of dmraid is that it's for handling hardware / fake software (i.e still on a card) arrays.

 

Hmm.. thanks... I guess I read it wrong

 

a `mdadm --stop` with a ubuntu live-cd might work according to the internet:

The --stop tells mdadm to remove its metadata (which is at the end of the partition). So then, because it is RAID1, it just becomes your normal, run of the mill, partition that can be mounted. (This would not work with any other RAID type, but in the case of RAID1, all the data is there.)

The theory is sound - though you're braver than me

 

Hope you have a backup

 

I'd be interested to hear how you get on with it.

 

If you can break the raid-1 array and have it normally mountable without the need for any kernel md drivers then the rest of it becomes much much easier - as I suspect you're fully aware!

though you're braver than me

 

I will just mark the hd as faulty and get it out, and let the other hd of the raid 1 intact while I try to remove the raid info on the removed one. In worst case, i `dd` clear the hd and add it back to the degraded array and let it rebuild

 

So not really a risky part in my eyes

I will just mark the hd as faulty and get it out, and let the other hd of the raid 1 intact while I try to remove the raid info on the removed one. In worst case, i `dd` clear the hd and add it back to the degraded array and let it rebuild

 

So not really a risky part in my eyes

 

Very good idea - I like it.

 

You're still braver than me though

Disconnect went good, the etrayz still has a working (degraded) raid 1 filesystem.

After some exciting minutes, i was able to remove the raid info from the other hd. and i can mount the partitions in the ubuntu live cd.

However (and I didn't check that before) the filesystem is a XFS system, and the unraid usb disk can't mount that by default.

 

Are there packages i can install so I can mount the XFS filesystem?

I would just do a long transfer across your LAN, probably take about 24 hours and then you're done.  That way you're doing nothing risky.

 

The unRAID end of things would be the usual bottleneck for transfers, typical writing speeds are in the 20-30MB/s range (with parity protection on), so about 24 hours is needed to write 2TB of data.

 

Is your device slower than 20MB/sec on reading?  If it was limited by a 100Mbit/s LAN then you wound need about 2 days to do the transfer.

 

Don't skip doing the preclear, a lot of us have had the preclear identify drive issues before they got entrusted with data.

 

Regards,

 

Stephen

unRAID does not support XFS. So you cannot mount it directly on unRAID.

 

 

Probably better doing an rsync across the lan.

 

 

Maybe Tom could compile XFS for you, but it's unproven to work with unRAID and you may need a mount command for the XFS filesystem.  I'm not sure it's worth the effort.

The main reason I wanted it out of the raid 1 array is so I can use the other disk as data disk.

I can now mount the xfs partition on my htpc (also an ubuntu (xbmc) machine) and do a (relative) quick transfer over the network (gbit).

I did have some content (400GB) on there aswel (as temporary storage, cause the raid array was full) to transfer.

 

Thanks for all the suggestions. It would taken me way longer to figure out these thing without the help and suggestions!

 

I'll start the preclearing of the remaining disk tomorrow

It all took longer than expected,

the hardware came in yesterday. But I had some trouble installing slackware on it.

Finally managed to get it running including the XFS filesystem. So I can do a local file copy

The system is now copying the data