I'm looking for some help on tracking down this problem I'm having. If I try to write two files at the same time to the server then it immediately crashes. I'm also looking for a way to be able to get a log of whats happening. Nothing works after the crash so I can't save the syslog. Can't log in when physically connected, nothing, have to power it down physically. Can write files sequentially but not simultaneously.  I can read files fine. Tried writing 2 files simultaneously to one disk or to multiple disks, all cause a crash. This is a newly built server, I suspect I have had this problem from the start.

 

Server specs:

CPU:	Intel Celeron 3.0GHz Single Core
MB: 	Gigabyte EP45-UD3LR
Mem: 	2x1GB  OCZ
PSU: 	Corsair TX650
Cards: 	2x SYBA 2 port SATA II (SiI3132) - Current not in use
Case: 	CM590
Caddys: 2x Cooler Master 4-in-3
HD's: 	2x 2TB WD EARS, 2x 500GB Seagate 7200.11, 1TB Hitachi, 200GB WD2000 (cache), No Parity
UnRAID: 4.5.3 Pro

 

Bios has AHCI on, and Native IDE for ports 0-3 (other option is Legacy).

 

Log of just before the crash is attached. If there is anymore info that would be helpful please let me know.

syslog-2010-06-22--1-13.txt

I just tried copying files simultaneously with no problems to a server from 3 computers: WindowsXP, Vista, and MacPro.

 

From your syslog, the cache disk is producing a strange error:

 

Jun 22 00:08:11 Tower kernel: attempt to access beyond end of device
Jun 22 00:08:11 Tower kernel: sdc1: rw=0, want=390721904, limit=237117440
Jun 22 00:08:11 Tower kernel: reiserfs_resize: can't read last block

 

What is the history of this drive?  Was it formatted using unRAID?

 

Also you should upgrade to 4.5.4.

 

When you say you write to one disk or to multiple disks, do you actually mean one share or multiple shares?  Or are you writing to disk shares?

 

If you are writing to user shares with cache disk enabled, and the above error is produced during cache disk mount, I'm not surprised there might be a crash.

 

To capture a syslog that might show the crash, open a telnet session and then type this command:

 

tail -f /var/log/syslog

 

Now try to make it fail.  Any generated syslog entries will appear in your telnet window where you can select/copy/paste into a post.

Upgraded to 4.5.4.

 

I pretty sure the drive was formatted with unRAID. The drive has 2 partitions. I'm going to reformat it and remove the second partition.

 

I was writing to the disk shares so the cache drive shouldn't have been involved.

 

I'll get that sorted out and try to get a log of the crash.

 

Thank you

There can not be two partitions if formatted via unRAID.

 

I tried same test as before, this time to disk shares & no problems.

Removed the cache drive. I don't think I get anything useful to show in the telnet window. Log is right before I cause the crash.

 

Sorry if this was posted in the wrong place. I haven't the slightest idea as to whats going on as I can't seem get any info of whats happening when the crash occurs.

 

Thank you for your help.

syslog-2010-06-22--4-27.txt

Absentmindedly I didn't try to cause the crash doing simultaneous transfer from a different computer. Turns out if I transfer files from a laptop running win 7 x64 it doesn't crash the server.

 

I'm running the same version of win 7 x64 on the computer causing the server to crash. My network controller is: Marvell Yukon 88E8053

 

What should I be using for settings in the advanced adapter settings panel?  I've try a couple settings but no change. What could be causing this?

 

Maybe this should be moved to hardware...?

I have not been able to find the reason why my computer can cause unraid to crash. I do know that other computers in the house don't cause it to crash. I am not blaming unraird for the crashes, it looks like it has something to do with my setup. BUT it does not seem right to me that I can cause unraid to crash with just writing multiple files simultaneous. I would hope that unraid would have some sort of protection agaisnt this.

 

I am willing to help figure out what causes this but would need some direction. I would also have to finish up building a testing box, I don't want to crash my unraid box with all my data on it.

 

What I do know:

Windows 7 64bit - Causes crash

Windows XP 32bit - Does not cause crash

Both run on the same computer (dual boot)

 

Windows 7 64bit on a different computer in the house does not cause crash.

This may not be a solution to your problem, but perhaps an acceptable work-around:

 

Get TeraCopy (it is free).  When you start multiple simultaneous transfers with TeraCopy installed, it will queue each one and wait for the previous to finish.  This also helps improve transfer speeds.  There are tons of other benefits as well.

 

I won't use any version of Windows without it, since the built in windows transfer dialogue is crap.

I actually had tried TeraCopy and SuperCopier and possibly a third option (can't remember now). I actually saw a decrease in speed using them. Which I know seems old but windows was actually quicker.

 

I also have a few programs that work with files on my unraid server running from my windows computer. TvReaname and Ember Media Manager, both those programs cause the server to crash also. I don't know what they use to send files to the server but it doesn't look like its the file transfer program in windows but something more low level.

 

I have sick beard running and moving files around within the unraid server with out any issue.

 

Its a pain not to be able to transfer 2 files at the same time. I have been just living with it. But the one time you forget and then you have to worry about causing damage to your data.

 

The part that is even more annoying is that even if teracopy was a work-around I wouldn't be able to use Ember Media Manager, which I need to use.

 

I do appreciate the suggestion Rajahal.

 

Hmm, it definitely doesn't make any sense that TeraCopy is slower than the Windows dialog.  Are you able to transfer multiple files simultaneously between your Windows computers (without unRAID involved)?

 

Also, do you trust all your network hardware?  Routers, switches, cables, NICs, etc?  Using Cat5e or better network cable?  Did you terminate any of the cable yourself?

If I get a chance I'll try teracopy again, and once I finish building a second unraid box I can test more.

 

I can transfer between my Windows computers with any issues.

 

Also, do you trust all your network hardware?  Routers, switches, cables, NICs, etc?  Using Cat5e or better network cable?  Did you terminate any of the cable yourself?

 

Using a switch between the unraid server and my computer. Any other comp in the house does not cause the unraid server to crash on simultaneous transfers. Cables are CAT6. I strongly believe it has to do with Windows 7 64bit and the NIC together. Using the exact same computer (dual booting) I can do simultaneous transfers in windows XP without issue.

 

So far according to the other thread, the people so far have reported being on Windows 7 64bit. Since not everyone on Windows 7 64bit is having the problem it must be Windows 7 64bit AND hardware, I assume the NIC. I have posted what NIC I have.

 

Hopefully we can find some relation between Windows 7 64bit and some NIC's. This is just me ASSUMPTION.

Do you have any WD advanced format drives without the jumper?  I had crashing issues, but things are much better now that I run all the EARS drives with the jumpers.

 

Note you can't just switch the jumpers - you will lose the contents of the drive.  You need to copy the data off, remove it from the array, and re-add it.

Do you have any WD advanced format drives without the jumper?   I had crashing issues, but things are much better now that I run all the EARS drives with the jumpers.

 

Note you can't just switch the jumpers - you will lose the contents of the drive.  You need to copy the data off, remove it from the array, and re-add it.

 

Or if you don't have enough free space for that, you can let unRAID rebuild the data for you by following this procedure.

Do you have any WD advanced format drives without the jumper?   I had crashing issues, but things are much better now that I run all the EARS drives with the jumpers.

 

Note you can't just switch the jumpers - you will lose the contents of the drive.  You need to copy the data off, remove it from the array, and re-add it.

 

All my EARS drives have had the jumpers on them since day one.

 

Like I have said before. Other computers in my house do not cause unraid to crash. The same computer that DOES cause unraid to crash does NOT crash unraid when in Windows XP.

Well I've been just dealing with the problem, by not doing multiple transfers. But it really bugs me and I'm hoping this can get fixed.

 

What I have just tried:

-Changing the SAMBA limit to 4000 (as outlined bellow)

-Tried to take a log by running this command: tail -f /var/log/syslog

but unfortunately I don't get any info before unRAID crashes.

-Disabled cache drive, not that it would have made a difference but tried anyhow

 

Should I try to change Linux's limit? My current limit is about 200,000

 

I really hate doing any testing on this server as its my server with all my files and crashing it over and over doesn't sit well with me.

 

One of the recent reasons for fixing this is XBMC wants to write some files to my server, when it does it crashes (since I assume its multi-threading of sorts). Previously I dealt with it by not allowing XBMC to write to the server. Same thing happens if I start 2 simultaneous on the same computer, the problem I've always had.

 

Also for some reason my smb-extra.conf is getting overwritten/blanked when I reboot. I do have SNAP installed and that might have something to do with it but I don't know why.

 

Related thread: http://lime-technology.com/forum/index.php?topic=8626.15

 

also i am using windows 7. ive used other oses before. same issue. if i have two file transfer windows open it does it. maybe the parity drive cant keep or something and crashes the system. i dunno.

 

also to clarify a little more, ive transfered large amounts of data before. hundreds of gbs at once. no issue. but if i copy some files to one directory on unraid and copy more files to unraid while the others are still copying, after a few minutes its screwed up. really inconvenient. i just avoid it because troubleshooting means waiting hours on end each time for parity check to finish.

It sounds more like you are reaching a limit on the number of "open" files.  Perhaps Windows7 is not closing the files after copying them?

 

Once the system wide limit on open file descriptors is reached, nothing is predictable... (and the server will probably soon crash if a critical process cannot open a new file.)

 

There are two different limits... one for Linux itself, the other in SAMBA.

 

In smb-extra.conf, in a "[global]" section, you might try:

 

[global]

max open files=40000

 

At the linux command line level, to see the current limit type:

cat /proc/sys/fs/file-max

 

To see what is used, type:

cat /proc/sys/fs/file-nr

 

For more info (and how to interpert the resulting output from the above command, look here:

http://www.netadmintools.com/art295.html

 

Solved in this thread:

http://lime-technology.com/forum/index.php?topic=16547.msg151951#msg151951