Stand-alone preclear box

[Kode]

I'm thinking of buying a really cheap atom or similar mini-itx board to use as a stand-alone preclear box, that way I can preclear with the server turned off, was thinking of maybe buying another http://www.overclockers.co.uk/showproduct.php?prodid=CA-055-XG as my server has 4 of them in it anyway, so I can mix and match the trays.

 

Does anyone know of any really small cases that also has 3 x 5.25 bays?

 

Also any other suggestions would be welcome.

 

I've bought an N54L to use primarily as a standalone preclear box but maybe also to test newer versions of unRaid as well.

 

I was also wondering if the preclear script relies on unraid for anything or is it self contained?  Because if it is self contained another possibility might be to automatically start the preclear process when a drive is inserted but unRaid doesn't support hot-swap (afaik) so it would need to be an OS that does support it.

[Kode]

Actually it might be easier / cheaper just to get something like http://www.ebuyer.com/430446-hp-proliant-g7-n54l-2-2ghz-microserver-ebuyer-704941-421 after cashback it's only £110 or maybe I can find an older/cheaper version on ebay if it's just for preclearing

[prostuff1]

I have a test server/ preclear box.  It has 6 internal drives for unRAID and then 6 trayless hotswap bays in the front.  I can clear up to 6 drives at once which is nice, seeing as I build servers for customers.  I can burn in larger capacity drive while I stress test the other hardware with smaller capacity drives.  I have a hoard of 160GB and 250GB SATA drives that get beat to hell when building servers.

[ironicbadger]

Or run a VM dedicated to it on an existing box? Seems like an awfully specific use case to justify a whole new box?

 

Fun though!

 

Sent from my Nexus 5 using Tapatalk

 

 

[Helmonder]

Or run a VM dedicated to it on an existing box? Seems like an awfully specific use case to justify a whole new box?

 

Fun though!

 

Sent from my Nexus 5 using Tapatalk

 

You would need to add a sata controller specifically pass-thru'd to that preclear vm though..

[ironicbadger]

Yes, could work out cheaper than a whole new box not to mention more efficient.

 

Sent from my Nexus 5 using Tapatalk

 

 

[prostuff1]

Yes, could work out cheaper than a whole new box not to mention more efficient.

 

Sent from my Nexus 5 using Tapatalk

I use my preclear box for more than just preclear.  The hardware in it is mostly hand-me-down/scavenged stuff that does not support VM's and passthrough.  Would I love to have another ESXi box, of course, but I have better things to spend money on then an ESXi test box.

[garycase]

I use an old PC that would have otherwise been donated to Goodwill as my pre-clear machine ... as I suspect most of us who have "dedicated pre-clear machines" do.

 

... so it was effectively free.  Only cost is the power to operate it.

 

[ironicbadger]

unRaid doesn't support hot-swap (afaik) so it would need to be an OS that does support it.

 

Yes it does with the right sata controller. My PCIe Supermicro MVSAS card works hotplug, but not via my motherboard sata ports.

[WeeboTech]

unRaid doesn't support hot-swap (afaik) so it would need to be an OS that does support it.

 

Yes it does with the right sata controller. My PCIe Supermicro MVSAS card works hotplug, but not via my motherboard sata ports.

 

 

The kernel may support hot swap for 'some' controllers.

The removable housing may support it safely at the hardware level for 'some' hardware.

The kernel may see it after a certain time.

For the purpose of preclearing, the hardware and kernel will pick up a new drive after it has been detected and the hardware supports it.

 

unRAID's array (md driver) does not support hot swap directly.

A person would have to stop the array and restart it in order to pick up a new drive.

[garycase]

... and regardless of the technical support for hot-swap (which any AHCI compliant controller is supposed to support as long as it's running in AHCI mode), it's still a good idea to shut down when changing disks.    Hot-swap capability is certainly convenient with OS's that support it ... e.g. for removable drives in Windows, where you can do a "Safely Remove" and then disconnect a drive that the OS has disconnected itself from.    I assume UnRAID does effectively this if you Stop the array and then change drives ... but it certainly seems better to just power down for the swap.

 

 

[ironicbadger]

What about the pre-clear script?

 

Will it run on other OS's or is it just unRAID?

[Kode]

What about the pre-clear script?

 

Will it run on other OS's or is it just unRAID?

I don't think so, it reads /var/local/emhttp/disks.ini

[ironicbadger]

Was afraid so

 

Sent from my Nexus 5 using Tapatalk

 

 

[WeeboTech]

It could be made to run anywhere. It's dd, awk and some fancy shell programming.

[Kode]

From the quick scan I took earlier, I think it's only reading the file to check which drives are not part of the array and so therefore candidates for preclear

[garycase]

Yes, it reads the UnRAID config to determine what drives are already part of the array.  But I don't believe that's a show-stopper -- if it can't find the info, you can simply pre-clear any disk (obviously you'd want to be CAREFUL about doing this ... although if it's a dedicated pre-clear box that's not really a problem).    I'm fairly sure you can run it on any Linux system ... but the simplest thing to do is just put a copy of the script on a USB flash drive with a basic UnRAID setup (no drives installed) => then just boot to that when you need to pre-clear.

 

[cal87]

I use my latest decommissioned desktop PC as a preclear box.

I have a couple 4TB drives clearing now.

 

I am planning on replacing some of the 2TB drives in my main array.

I was thinking about using the same preclear box to copy data from the existing drives to the new ones.

After I am done, place the new drives in the main array, shrink the array, and then do a parity sync.

Any reason I should not do this?

[JonathanM]

I use my latest decommissioned desktop PC as a preclear box.

I have a couple 4TB drives clearing now.

 

I am planning on replacing some of the 2TB drives in my main array.

I was thinking about using the same preclear box to copy data from the existing drives to the new ones.

After I am done, place the new drives in the main array, shrink the array, and then do a parity sync.

Any reason I should not do this?

Sounds like a good plan, especially since you will have copies of the data you are manipulating, so risk is reduced.

The only thing I can add is to use a checksum verification on the old and new data, just to be sure. Either use a checksumming copy program like teracopy, or create MD5 files before you start and verify after copy.

 

Do you already have a 4TB parity drive in place, or is one of the currently preclearing drives going in that slot?

[cal87]

Sounds like a good plan, especially since you will have copies of the data you are manipulating, so risk is reduced.

The only thing I can add is to use a checksum verification on the old and new data, just to be sure. Either use a checksumming copy program like teracopy, or create MD5 files before you start and verify after copy.

 

Do you already have a 4TB parity drive in place, or is one of the currently preclearing drives going in that slot?

 

Already have a 4TB as parity and very recently did a full parity check.

[garycase]

Copying the data in your spare box; then doing a New Config and including the larger drives in it will work fine -- but you'll be "running at risk" for the duration of the initial parity sync.

 

Not a big deal ... but just something to be aware of.

 

You can do this without running at risk by simply replacing, one-at-a-time, the 2TB drives in the array with a 4TB drive, and waiting for the rebuild to complete.    As long as you don't do any writes to the array, you'll still be protected in case of a failure, as you could simply replace the 2TB drive and do a New Config with "Trust Parity" checked to restore the system back to its previous state (and then replace whatever drive had failed in the process).

 

But I agree that as long as you KNOW you're starting out with everything good (i.e. you just ran a correcting parity check and there were zero errors and zero sync corrections), that the risk of just doing as you noted is very small.

 

[cal87]

Copying the data in your spare box; then doing a New Config and including the larger drives in it will work fine -- but you'll be "running at risk" for the duration of the initial parity sync.

 

Not a big deal ... but just something to be aware of.

 

You can do this without running at risk by simply replacing, one-at-a-time, the 2TB drives in the array with a 4TB drive, and waiting for the rebuild to complete.    As long as you don't do any writes to the array, you'll still be protected in case of a failure, as you could simply replace the 2TB drive and do a New Config with "Trust Parity" checked to restore the system back to its previous state (and then replace whatever drive had failed in the process).

 

But I agree that as long as you KNOW you're starting out with everything good (i.e. you just ran a correcting parity check and there were zero errors and zero sync corrections), that the risk of just doing as you noted is very small.

 

Took me a while thinking about it.  If there is a failure DURING that parity sync, I could be in trouble.  I could mitigate that by copying the parity drive (not sure the fastest way to do that, but I keep an extra 4TB as a hot spare).  Does that sound right?

 

Think I will stick with the original plan and take the small risk.

[garycase]

No need to copy the parity drive -- just do the one-at-a-time drive upgrade and let them rebuild.    As long as you don't write to the array during those rebuilds, you can always replace the original drive and do a New Config with "Trust Parity" to get right back to where you started.

 

That protects you against a failure during the upgrades.    Not likely ... but nice to know it wouldn't cause any problems if it did happen.

 

[cal87]

No need to copy the parity drive -- just do the one-at-a-time drive upgrade and let them rebuild.    As long as you don't write to the array during those rebuilds, you can always replace the original drive and do a New Config with "Trust Parity" to get right back to where you started.

 

That protects you against a failure during the upgrades.    Not likely ... but nice to know it wouldn't cause any problems if it did happen.

 

Understood.  Just trying to figure out a faster way to do it then one by one, since I will also be combining data from 2 drives into one.  I should just do it instead of thinking about it too much.  I'm starting to chicken out.

[garycase]

Just do the drive replacement one-at-a-time ... you're only talking about 2 drives, right?

 

That's faster than copying the data; then doing a New Config and parity sync ... and you'll never be running "at risk"  (as long as you don't write to the array during the rebuilds, as I outlined earlier).

 

Once they're both done, if you want to you could move data from one drive to another ... but there's really no reason to do that.