HI all,

 

Not too rehash duplicate threads or problem. But I have been getting sync errors lately of which I never use to get any. I have tried duplicated parity checks one right after another and the errors do decrease they are still present. I have both servers on a APC 900 and I am running 4.3.3 final. I have no stop, sync power off or powerdown commands available on my system so when the power goes I have to use a shutdown -h now command.

 

Then I was going to start a new parity check last night around 8PM so It would be finished this morning and to my surprise when I loaded the monitor window, the server had started a parity check on it;s own from a running array.

 

it is just finishing up the check now and found 122 errors this time around. I had a drive fail last week (it was a seagate 1Tb only a month old and it would not boot period) I rebuilt the drive and ran the check on that disks replacement and that is when the errors seem to start.

 

This parity check starting on it's own has me concerned. I will post a log as soon as this check stops.

 

When running the smart ctl short do you need to stop the array?

 

THX,

 

Dave

 

 

When running the smart ctl short do you need to stop the array?

No.

Thanks Joe,

 

Here is a link  to the log. it is 2100Kb which is quite large for a log, so I could not post it here.

 

http://www.scottcraftboats.com/syslog.htm

 

Another note about this, I switched out 5 of the Hd caddies for a Supermicro 5-3 chassis. Which is you are not familiar with does have 5 sata connectors (data) and is powered only by two OR three molex connectors. I am wondering if this thing is causing some of my problems as well.

 

thanks,

 

Dave

Thanks Joe,

 

Here is a link  to the log. it is 2100Kb which is quite large for a log, so I could not post it here.

 

http://www.scottcraftboats.com/syslog.htm

 

Another note about this, I switched out 5 of the Hd caddies for a Supermicro 5-3 chassis. Which is you are not familiar with does have 5 sata connectors (data) and is powered only by two OR three molex connectors. I am wondering if this thing is causing some of my problems as well.

 

thanks,

 

Dave

Your log shows many warnings about the file-system structure on /dev/md2.  This needs to have a reiserfsck run on it.  I'm certain the file-system is corrupted and needs repair.

 

Basic instructions are i the wiki: http://lime-technology.com/wiki/index.php?title=Check_Disk_Filesystems

 

Joe L.

Hey Joe,

 

ran it and it skipped 5 of 6 nodes on initial and then said the bitmaps did not match. Disk2 was the disk that failed on me a couple weeks back. I am running a --rebuild-tree right now.

 

dave

Well it finished the rebuild tree and placed two complete video files in the lost and found folder. Which is weird, these are probably a couple of the firist movies I ever placed on the drive and I have not watched them in quite a while. Oh well I guess I will re-rip the two of them and see what happens.

 

Thanks for the help joe. I reran the ck after the rebuild-tree was complete and it found no more corruption. So Hopefully things are back to normal.

 

rgds,

 

Dave

 

I think I have found some further problems. When I did the rebuild tree the system created the lost and found folder and at best I could find a couple files that it placed in that folder. I replaced those files and deleted the lost and found folder. Last night I went view a few more files to find there was no data in the directories at last count it was about 12 files in separate directories.

 

If I view the folders within windows I see nothing in the folders for data, but if I try to delete them windows gives me an error saying the folders are not empty.

 

What is the command to remove these from a telnet window or the like. The problem is still on disk 2.

 

Thanks,

 

Dave

 

I think I have found some further problems. When I did the rebuild tree the system created the lost and found folder and at best I could find a couple files that it placed in that folder. I replaced those files and deleted the lost and found folder. Last night I went view a few more files to find there was no data in the directories at last count it was about 12 files in separate directories.

 

If I view the folders within windows I see nothing in the folders for data, but if I try to delete them windows gives me an error saying the folders are not empty.

 

What is the command to remove these from a telnet window or the like. The problem is still on disk 2.

 

Thanks,

 

Dave

Easiest from telnet, if you are not familiar with Linux commands, is to type

mc

and invoke midnight commander. 

 

From there you should be able to see the files and delete them.

 

Other than that, you can use:

ls -al

to list all the files in a directory, even those with a leading "." that are normally hidden

 

and

rm filename

to remove a file

and

rmdir directoryname

to remove a directory (note:, it must be empty to remove it)

 

Also note that in Windows, unless you set the options to "see hidden files" and "show system files" in file-explorer, you may not see some of the files present in unRAID folders.

 

tried those commands as well. I get an input output error when trying to read any of the affected folders.

 

root@Tower:/mnt/disk2/Movies2/Rambo# ls -al
/bin/ls: reading directory .: Input/output error
total 0

 

Dave

 

tried those commands as well. I get an input output error when trying to read any of the affected folders.

 

root@Tower:/mnt/disk2/Movies2/Rambo# ls -al
/bin/ls: reading directory .: Input/output error
total 0

 

Dave

Then it is probably time for you do to another reiserfsck to check for corruption.

 

Joe L.

If I view the folders within windows I see nothing in the folders for data, but if I try to delete them windows gives me an error saying the folders are not empty.

 

Did you try turning on the options to let you see hidden and system files? 

 

To do this in XP, from Windows Explorer, Tools Menu, Folder Options ..., View Tab

-  Select - "Show Hidden Files and Folders"

-  DEselect "Hide protected operations system files").   

 

After running reiserfsck to rebuild my entire tree, it set all the attributes to something that made seeing them from the windows side problematic.

 

I also needed to reboot my unRAID server before I could see anything.

I just ran reiserfsck and yes it asked for me to rebuild the tree again. What is causing this, this is the 2nd time in less than two weeks I have had to do this.  It is a brand new drive.

 

All my windows machines have those setting turned on and off so I can see hidden files. I am also pretty sure I rebooted the system following the rebuild.

 

thx,

 

Dave

Have you performed any SMART tests on the drive?, the additional errors could very well be caused by it being a "new" drive. (they are more prone to failure) 

 

You should fix the corruption as instructed by reiserfsck, then run both a "short" and a "long" smart test on it.

 

Details here in the wiki: http://lime-technology.com/wiki/index.php?title=Troubleshooting#Hard_drive_failures

Joe,

 

I ran both tests, I cannot get it to put the report on the flash for some reason. should it be in the root, I have not shut down the server and pulled the flash to check, I just assumed it would be in the root.

 

Thanks,

 

Dave

Joe,

 

I ran both tests, I cannot get it to put the report on the flash for some reason. should it be in the root, I have not shut down the server and pulled the flash to check, I just assumed it would be in the root.

 

Thanks,

 

Dave

There is no "output" when requesting a long or short test... The disk itself runs the test and store the test results internally on itself. The output is subsequently requested by using:

smartctl  -a  -d  ata  /dev/sda >/boot/smart.txt

or

smartctl  -a  -d  ata  /dev/sda | todos >/boot/smart.txt

(The second format makes it easier to look at the smart.txt file from a Windows workstation)

 

Then, the file it creates would be seen at \\tower\flash\smart.txt    Long tests take several hours, short tests several minutes. You need to wait before you will see their results. 

 

(Obviously, in the examples above, change /dev/sda to represent the disk you are having trouble with.)

 

Hi Joe,

 

Yep! I understand all of that, what I am saying is that neither addt'nl command produced a report to view in the root of the flash drive. This was with a short or long test.

 

So I started the short test with the proper command line after the smartctl time stated to complete the test I issued the report command and nothing has shown.

 

Dave

figured it out. I forgot that I had reinstalled windows a couple weeks ago on this machine and had not removed the check box for system files. OOOPs.

 

So I performed a long test last evening on the drive if I create a report now from smartctl. Will it give me the report from the long test now.

 

Thanks,

 

Dave

OK! so I think I finally figure out this reporting deal. it writes the report for any test you have done with a smart.txt.

 

here is the report, I cannot see anything major wrong. I am almost afraid to put any data on this drive, the last time I ran rebuild tree it found only a couple files and this time it found 17 orphans I guess you would call them.