Ok, what's going on?

I've been trying to preclear a 2TB disk and I've tried 3 times now, and each time it gets to 5% on the writing zeros stage and then stops and just says "Killed"

 

Is the HDD kaput? Even though it's brand new?

Why "Killed"?

Are you running the preclear in screen or straight from the login session? And if straight from the login session, are you running with nohup? It's possible it just happens to die after 5% because the session that started the process ended.

I'm running it from the login session plugging a screen straight into the server.

I don't know what nohup is though.

 

But it has worked on my other 2TB drive, but with this HDD it just keeps dying at 5% and says "Killed"

No I mean through screen - see here. Screen is like a multi-shell window manager, but also has the benefit of not dying once you've logged off.

 

Are you leaving it to just run, or do you close stuff and do other things while it's running?

 

Killed usually means the process was explicitly killed by something. I guess a syslog would help...

I'm not using telnet or screen, I'm just plugging in a monitor to the VGA port on the server itself, and I'm logging in and running the preclear_disk.sh script,

I do disconnect the monitor for use on a different machine whilst it runs, but this didn't affect the previous HDD preclear, but this time something not working.

Please describe your hardware configuration and Post your syslog.

Usually when it says killed, it means there is an out of memory condition.

There is only 512mb of memory, if you think it's that, but it worked for the other 2TB drive, so why not this one?

I don't know what else is running in your machine, how many file/directories are cached and what other activity is going on.

It reads as if there's an out of memory condition.

We can't help without a syslog.

Ok, well I've restarted the machine since then so the only system log I can find at http://Tower is of 2 minutes ago when i turned it back on.

cp /var/log/syslog /boot/logs

Run the pre-clear until it is killed and then collect the syslog.

ok, well I'll run the pre clear again just incase it decides to work this time, and if/when it doesn't I'll have the syslog for you.

If you want to just gather the latest data while the preclear is running:

 

tail -f /var/log/syslog | tee /boot/logs/syslog.txt

 

Just in case you get really flooded logs or something. Having said that, I'm sure we're all capable of scrolling.

nope thats not working, using

 

tail -f /var/log/syslog | tee /boot/logs/syslog.txt

 

through telnet I'm just getting a record starting from 18:25 which is only 15 mins ago, but the pre clear is running again anyway, so I should either have a cleared disk or a syslog to send by tomorrow.

Since you've recently rebooted, chances are the preclear may end successfully.

I would run as little as possible on the machine or refrain from access until the preclear finishes.

Ok, well I've just checked and it's done it again, 5% and it's just stopped and says KILLED, but I can't connect to it to get a syslog, either through the cpanel at http://tower or with telnet through terminal (as I have a mac), it just says "unable to connect to remote host" so I'm not sure how to get a syslog to show you guys.

I agree with previous suggestions - sounds like the OOM killer is getting involved.

 

What other plugins etc are you running on that machine?

 

Can you open a terminal whilst a preclear is running and keep an eye on the output of free -m?

 

Either you have too little memory to do a preclear (unlikely?) or you have something else that is eating memory.

I have no other plugins as far as I know, I just downloaded the basic unraid package, then decided to wipe these disks so downloaded preclear to the USB stick and that's it.

Nothing else on there, and 512mb memory isn't much, so if it hadn't worked fine for the other 2TB drive I would be inclined to think it was that.

I have no other plugins as far as I know, I just downloaded the basic unraid package, then decided to wipe these disks so downloaded preclear to the USB stick and that's it.

Nothing else on there, and 512mb memory isn't much, so if it hadn't worked fine for the other 2TB drive I would be inclined to think it was that.

 

Did you try preclearing without installing ANY plugins?

What do you mean? I don't have any plugins, I did exactly what I said, extracted the unraid server folder to the USB stick, then downloaded the preclear script and added that to the USB stick and thats it.

I don't know what plugins are, so I doubt I've installed any.

What do you mean? I don't have any plugins, I did exactly what I said, extracted the unraid server folder to the USB stick, then downloaded the preclear script and added that to the USB stick and thats it.

I don't know what plugins are, so I doubt I've installed any.

how much RAM do you have installed?

 

What is the output of the following two commands:

free

df

 

Have you run a memory test? (several cycles, preferably overnight)

I have no other plugins as far as I know, I just downloaded the basic unraid package, then decided to wipe these disks so downloaded preclear to the USB stick and that's it.

Nothing else on there, and 512mb memory isn't much, so if it hadn't worked fine for the other 2TB drive I would be inclined to think it was that.

 

When linux runs out of memory it starts killing off processes to bring usage back under control.

 

This would explain why your preclear is sent the kill signal and why your web control is also terminated (emhttp is usually a good candidate for being killed off too).

 

It all fits, you need to monitor your memory. As unraid lives as a RAM disk it could even be something like a runaway log file filling up your 'disk' (which is actually your ram).

 

Doesn't explain why you could previously clear another 2TB disk unless something has changed. Perhaps the disk you're preclearing is spewing log errors out and chewing up memory.

 

Out the box unraid shouldn't use too much memory, certainly less than 512 so you'll need to dip in and have a look to see what's happening. Or don't preclear the drive and let unraid do it for you 'normally'.

What do you mean? I don't have any plugins, I did exactly what I said, extracted the unraid server folder to the USB stick, then downloaded the preclear script and added that to the USB stick and thats it.

I don't know what plugins are, so I doubt I've installed any.

 

So sorry, I'm confusing this with another user who is having out of memory issues also.

In any case, it reads like there is memory pressure. I'm not sure 512MB is enough any more.

I think the basic ramdisk starts to take up at least 256MB or more.

 

issue a du -hsx / and report back. on mine it is 414M.

 

You can also try the following after a freshly booted system.

 

sync

echo 3 > /proc/sys/vm/drop_caches

sysctl -w vm.vfs_cache_pressure=200

 

Then do the preclear on the new disk, do not use the system while this is occurring.

I suppose you could stop the array and free any memory that the md driver may be holding for stripes.

 

If the preclear succeeds, then it's memory pressure.

Reboot after the preclear to bring everything back to normal.