I have 10 hard drives that are 4TB and 3TB that need to be precleared and the same on an exact copy of my server belonging to a friend, How many hard drives can I preclear in one go and is there a limit using the unRAID free, 3 drive version. Im using a mac and Im trying to get my head around it but its a slow process, I just need to wipe all 20 drives on two servers, some have been running for over 1 year and the others for about 7 months so should I just do a 2 pass wipe to be safe or will one be enough.

 

Thanks

I believe 6, but I wouldn't do more than three.  My Supermicro controller would over heat if I did all 6 causing me to restart the preclears.

Okay thanks for that, also I understand you need to use unmenu and screen, in case the telnet windows close but do I need to worry about that as I will be logging into the server using the KVM on my super micro motherboard from my macbook pro and running it directly, will it matter if I then close down the KVM java window on my mac once it is all running or can I only open one preclear window this way. I ust need to know the easiest way to run it without monitoring as the 4TB drive are going to take a few days to preclear.

 

Thanks

If you use KVM, you don't have to worry about the process terminating when you close the window, but you'll still need screen to run more than one instance. Macs have telnet built-in anyway, so that's probably better than messing with a Java-based KVM. Just open terminal and type

telnet ip_or_hostname 23

 

Just make sure each preclear is in its own screen session and you can close the window. So, type "screen", start first preclear, type "CTRL-a c", start second preclear, and repeat. "CTRL-a n" will cycle through the windows. When you reconnect, "screen -x" will reconnect the last screen session.

Thanks I will do that, I found a way to install screen with unmenu so I will run it from terminal as you suggest.

One question though if I login to my server using java kvm and follow these guidelines as in unRAID pre clear user guide:

 

"If you did not install screen, you can preclear your hard drives from the console. Linux provides 6 Virtual Terminals (V-Terms) that you can use from the console".

 

Does this mean if I close the kvm the v-terms will close or not, i.e. will it avoid me having to use screen.

 

Thanks

If you use KVM, you don't have to worry about the process terminating when you close the window, but you'll still need screen to run more than one instance.

No - no need to use Screen.  Just use Alt-F1 through Alt-F6 to access up to six concurrent log-ins.  Works with Preclear just as with anything else.  And it works whether you use a KVM or not.

That should work. I'd test it to make sure first. I believe it's ALT-F1 through ALT-F6. Just type something into one of the virtual terminals, disconnect, reconnect, and make sure it's still there.

 

EDIT: oops, S80 beat me to it.

That should work. I'd test it to make sure first. I believe it's ALT-F1 through ALT-F6. Just type something into one of the virtual terminals, disconnect, reconnect, and make sure it's still there.

 

EDIT: oops, S80 beat me to it.

 

Using ALT+F1 to F6 also works using a keyboard and monitor hooked right up to the server, that's how I run my preclears.  This way you don't need extra plugins installed in order to keep the preclears running after closing telnet on another computer.

If you want to run using the console (alt-F1 - alt-F6), the limit is 6.

 

If you want to run using screen, there is no real limit. You can do as many as you have ports to connect drives. I've never had heat issues but I've never run more than about 4 at a time.

 

For most people with existing arrays, the number of new drives they are wanting to preclear at one time is usually a pretty small number (1 or 2, maybe 4 at the most).

 

Just remember not to run them in an open telnet session. When the session times out due to inactivity, or the window gets closed, your preclear stops.

Using screen, you can also split windows as many ways as you'd like.

 

Thanks for all of the useful information guys, I will try the Alt keystrokes for new windows tomorrow when I preclear the first set of drives, I do think I will limit it to 4 drives at a time, I don't want the LSI 9201-8i cards on my friends server overheating.

I do think I will limit it to 4 drives at a time, I don't want the LSI 9201-8i cards on my friends server overheating.

Preclear is no more stressful than a parity check on the drive and controller subsystem. The limiting factor seems to be memory, and low memory at that. So... limiting the number of simultaneous preclears may be necessary if you don't tweak the stock preclear values to reduce RAM usage, but overheating the HBA's should not be a concern. If it is a problem, then running unraid (or any multi drive raid setup) is going to be a problem on that machine.

Thanks for that input Johnathanm, Im running 16GB ECC ram and was going to run 5 preclears in one go, two time for each drive. I found an example of using the command on a forum topic but I can't find it now, do you thing I will need to worry with that amount of ram. Im struggling to find the correct command to say clear drive sda twice over and use a memory allocation in the command if need be, I know I don't need the -A command as all the drives are 3Tb and 4TB drives. Im using a LSI 9201-16i and my friend has 3 x LSI 9211-8i HBA cards.

I only ever run preclear one cycle. Every disk that I have ever precleared and came out clean has been long lived. And when you think about reading the entire disk twice and writing to the entire drive once you're probably simulating writes equivalent to years (close to a lifetime for many unRaid data disks) of use. How much life are you removing from the drive in search of increased reliability?

 

I researched this a few years ago and the conclusion from the commercial world was that very lengthy burnin did more harm than good, and the value of even a lighter pass was negated by the increased labor cost and having the drive in a RAID array (which will eventually kick a bad disk). But I still like to run the one full cycle as it does find the very occasional bad drive which I'd rather exchange for a new one than return via RMA for a refurb.

 

I'm sure others will chime in in the benefits of multi-pass but this works for me and doesn't take too long either. My opinion is to buy good drives (see the Backblaze study) and preclear once. Works better than buying a piece of crap (one Seagate model (2T?) had a 28% failure rate per year compared to a similar Hitachi that was under 2%) and testing it endlessly.

I also run one preclear per drive.

 

I have precleared 3x 4TB drives at once, but it can use a fair bit of RAM. I think I did it on my spare box that had nothing else running.

 

Sent from my Moto G using Tapatalk.

 

 

I run 3.  I've had drives DIE on the 2nd (~20%) and 3rd pass (~10%) before but most have died either before ever doing any (~40%) or after the 1st pass (~30%).  Don't feel it is worth it to go more on a new one but I do 6 on refurbished drives I get back from RMA since often they end up with 30-90 day warranty - whatever was left on old drive. 

 

Depending on the drive I have DOA of 0% (Hitachi) although 2% due to old age (one still going strong at ~30-40,000hrs), >50% WD Green after the 1st 20 5-10% in first 20 haven't bought one in couple of years, 0% WD Red 2TB, ~5-10% WD Red 3TB, ~10-15% WD Red 4TB, <10% ST3000DM001 Seagates but they die > 50% just before or just after warranty expires will not buy another but might try NAS drives, <1% Toshiba 3TB (but have only bought 13 so far - 1 bad after a year or so).  Some of the above may have been due to faulty newegg packaging (up until recently) but I tried to factor that in and reduced the %s a little on the WD Red and Seagate drives and I probably should have reduced the WD Green % as well but didn't.

Thanks Guys.

Neilt, how much ram did you have, I will run 1 preclear on 5 x 3TB drives on one server and 1 preclear on 2 x 4TB and 3 x 3TB drives on the other as per advise from you guys. I have 16GB ECC ram in each server.

You should not have any problem running 5 preclears with 16G of RAM. And ECC is fine but doesn't make it run more preclears - just helps to ensure you don't have memory errors get passed to the OS.

 

Good luck!

Just a thought

• If running under unRAID v6 (which is 64-bit) then all the 16GB of RAM should be available for use without any restrictions.  Note also that if using v6 you need the latest pre-clear script as earlier versions appeared to complete OK. but failed to set the pre-clear signature correctly for large disks on 64-bit systems.
  
• If running under unRAID v5 (which is 32-bit) then you are restricted to 4GB of 'low' memory.  I think the pre-clear only uses the 'low' memory so that may limit the number of pre-clear sessions you can run without encountering RAM issues.

Thanks Guys.

Neilt, how much ram did you have, I will run 1 preclear on 5 x 3TB drives on one server and 1 preclear on 2 x 4TB and 3 x 3TB drives on the other as per advise from you guys. I have 16GB ECC ram in each server.

3GB or 8GB, but using uRAID 5.0, so you are limited to lowem. Which I think is way less than 4GB. Which is why I didn't run anything else.

I've run 6 with default settings on a server with 4G RAM and minimal add-ons. I used ALT-F1..F6. I think I was limited to 4 in 2G RAM.

I have both servers running with 5 drives using alt-F1 etc and its running fine so thanks for all your help guys.