Skip to post #7 for the new lockup issue.

 

So I bought unraid pro, put a bunch of drives in and away we went - things were working pretty well. Long story short, I've transported the server (which seems to have affected a memory stick), taken some drives out to take pics of serial #s etc and put them back in the wrong way. The array was stopped (obviously) so I shut it all down, plugged the drives in correctly and powered back up.

 

All green status, great! Spoke too soon...

 

The parity drive and disk 1 were showing activity. I glanced over to disk 1's free space and it said "mounting"!!! Reads/writes to the drive climbing. I immediately powered down as that's the only drive that has data - a fair amount of data on it.

 

Why would it be doing this without any input on my part? Any ideas on how to correct it?

 

Thanks,

 

TyantA

Whenever a ReiserFS disk is mounted, there will be a certain minor amount of writes done to it. A small amount is normal.

 

"Mounting" means the system is in process of 'mounting' the drive, or making it available to be used. For more information see: http://whatis.techtarget.com/definition/0,,sid9_gci214638,00.html

Sometimes if the server is not shutdown correctly, the journal to a drive is replayed and this could be the delay in mounting and the writes occurring to the drive.

 

I don't know if swapping the drives would cause a rebuild of a drive like that without some intervention.

When you bring up the server again, copy the syslog to your flash and upload it so we can see.

 

or you can access your syslog with  http://tower/log/syslog

Sometimes if the server is not shutdown correctly, the journal to a drive is replayed and this could be the delay in mounting and the writes occurring to the drive.

 

I don't know if swapping the drives would cause a rebuild of a drive like that without some intervention.

When you bring up the server again, copy the syslog to your flash and upload it so we can see.

 

or you can access your syslog with  http://tower/log/syslog

Dependig on how much activity was occurring just before the unRAID array is stopped, especially if it is stopped abruptly. without the array being in a "Stopped" state first, it can take several minutes for the journal transactions to be re-played in the file-system.  These have been observed in large drive to take as much as 5 or 6 minutes, during which you'll see "mounting"

 

Joe L.

Wow, thanks for the prompt replies! I'll turn it back on then and check out the syslog. I was just afraid of data being overwritten seeing as both parity & disk 1 were reading /writing. Also, none of the shares were showing up either.

 

Edit: there's one detail I left out. While the drives were out, I was noticing two of them were not spinning up (nor being detected) by the controller. I forgot that some time ago, when these drives were in a raid 5 array, I used the "staggered start" feature of the highpoint rocketraid controller they were then attached to. Before making this realization, I plugged "disk1" into my main windows rig just to see if I could get it to spin up over there. The drive was detected but would not spin up. In computer management, the disk was detected and it asked if I wanted it to be initialized. I cancelled and shut down the computer, removed the drive as I realized what was going on and put it back in the server.

 

Everything seems to be functioning normally, just two drives that had the staggering "feature" applied, wait until quite a bit later in the booting process before being recognized. I wonder if this has to do with why I was getting sync errors previously. Anyway, that's another topic.

 

I've started the server and will let the mounting process run its course. Would you look at that. This time it showed up just fine. All the shares are back too. False alarm!

 

 

just two drives that had the staggering "feature" applied, wait until quite a bit later in the booting process before being recognized. I wonder if this has to do with why I was getting sync errors previously.

 

The Highpoint controller probably enabled power up in standby on the drive's software.

There is an hdparm setting to enable / disable this setting.

There are also jumpers on the later WD drives too.

 

The power up in standby and sync errors have nothing to do with one another.

Interesting. When I applied the setting, I received a warning saying it would not be reversible. I've searched Google a few times (not really know what I was looking for) to see if I could reverse it as I no longer find it necessary with a bigger power supply.

 

So maybe I spoke too soon. Everything showed up but I'm not having the greatest time copying any data from the drive. It'll transfer a few MB then lock up. Time to get that syslog posted...

 

Edit - I managed to get 160MB total off the server and it locked up completely. I rebooted and it seems to have decided it needs a parity check so that's what it's doing. I've now attached the syslog since the last reboot.

 

Edit 2 - I can only ever get a few hundred MB off at a time. I'm running memtest on the system now to see if that tells any tales. What's even stranger is I took the drive in question out of this unraid box and suck it in another system which has been tested to run unraid fine... and it exhibits the same issues. It'll copy a few hundred MB off then completely lock up. I'm forced to reboot. I'm guessing my issue(s) are related to the fact that I'm using a WD5000AAKS drive. I've had no issues with my WD6400AAKS drives so I thought I might be fine with the 500GB models. It appears I thought wrong. 

 

Any ideas on how to troubleshoot this!? I really just need to get the data off this drive.

syslog_4-30-2010.txt

Interesting to note, based on some info I've read in the forums, that when the server appears to be locked up on one system, I can still access it from another. I'm still not sure what the means but it appears to be a valuable clue. I'm tempted to just take the drive out and figure out how to get it properly recognized in my windows box to pull the files off it. I'm not concerned with getting these drives to play nice - they're not going back into the unraid box. I just need to get my data off this one drive.

 

Plugging a single data drive into a fresh unraid install still causes lockups as well. Grr.

The Highpoint controller probably enabled power up in standby on the drive's software.

There is an hdparm setting to enable / disable this setting.

There are also jumpers on the later WD drives too.

I always thought that disks don't remembet that setting after power cycle, hence the need for a jumper for that.

I'll have to look into that.  Power up in standby may be a useful feature.

 

The drives definitely do remember that feature, at least when set by my highpoint rocketraid controller. I have since formatted the drives, moved the drives to different controllers etc and they still won't spin up until prompted.

I have since formatted the drives, moved the drives to different controllers etc and they still won't spin up until prompted.

Is that creating some problems?  I'd think one would like it that way.

I mean, you have some problems since you are posting here, but your problems may not be related to this feature.

 

I have since formatted the drives, moved the drives to different controllers etc and they still won't spin up until prompted.

Is that creating some problems?  I'd think one would like it that way.

I mean, you have some problems since you are posting here, but your problems may not be related to this feature.

 

 

I do not think the feature is related to the problems I'm having. I wasn't the biggest fan before I moved to unraid as my Riad 5 array was in a desktop. Having each drive spin up individually - one after the other - meant waiting quite some time before the system booted. In an unraid box that's running most of the time anyway, sure.

The drives definitely do remember that feature ... they still won't spin up until prompted.

Are you sure?

Let's do a little experiment. (nothing your disks will remember)

1. Stop your server from the unRAID Management web page.

2. Telnet to your server, and issue the following command:

echo 'logger "###disk_status###" ; for i in /dev/[sh]d? ; do logger "`hdparm -C $i 2>/dev/null`" ; done' >> /boot/config/network.cfg

(You better copy from here and paste it in the telnet window, so you don't make a typing mistake.)

3. Reboot your server from the unRAID Management web page.

4. When the server boots, get the syslog and post it here.

 

The above will log the disk status to syslog early in the boot process, before any disk mounting has started.