March 3, 200818 yr I have two unRAID servers and recently one of them needs to perform a parity sync every time it starts. Yes, it even does so on a proper shutdown or reboot. For example, I just went into the web interface and stopped the array and the selected reboot and when it returned the parity drive had an orange indicator and the array was performing a parity sync. Why would this be happening? P.S. I'm running a pro key on version 4.2.1
March 3, 200818 yr I had the exact same problem. I posted the way I fixed it. http://lime-technology.com/forum/index.php?topic=1439.0 I hope this helps you. Phil
March 3, 200818 yr It would do that if it was unable to write the array status correctly to the super.dat file on the flash drive after it has shut down the array. Does it have a "write protect" switch (some drives have a physical switch to make them read only) If not, odds are the usb flash drive is starting to act up on you. Type: ls -l /boot/config/super.dat When was it last updated? It should be when the last clean shutdown occurred, or when the last complete parity calc completed. My super.dat shows this: ls -l /boot/config/super.dat -rwxr-xr-x 1 root root 4096 Mar 2 18:37 /boot/config/super.dat* My web-management page shows that parity was last completed as follows: Last checked on 3/2/2008 6:37:54 PM, finding 0 errors. Joe L.
March 3, 200818 yr Author /bin/ls: /boot/flash/super.dat: No such file or directory that's what is returned after running your command Joe L.
March 3, 200818 yr Author It seems like you have two different syntax for that command in your post Joe L. You say to type: ls -l /boot/flash/super.dat That returns: /bin/ls: /boot/flash/super.dat: No such file or directory Then I notice in your example you seem to enter: ls -l /boot/config/super.dat When I try that I get: -rwxr-xr-x 1 root root 4096 Mar 2 20:16 /boot/config/super.dat* That seems right about the time I properly shutdown the server.
March 3, 200818 yr Do you have any drive letters assigned to the unraid from your PC? I have my itunes folder on my unraid with an allocated device letter assigned to it as well as another for scripted backup. I periodically need to stop & restart the server in order for the user shares to be updated and I find that it never shuts down cleanly, on restart a parity rebuild is initiated. If I stop this parity rebuild & restart again the parity appears just fine again. Have you tried stopping that parity rebuild and restarting the array again? Mark.
March 3, 200818 yr It seems like you have two different syntax for that command in your post Joe L. You say to type: ls -l /boot/flash/super.dat That returns: /bin/ls: /boot/flash/super.dat: No such file or directory Sorry... of course you are correct. the correct command is ls -l /boot/config/super.dat
March 3, 200818 yr Author I take it the data returned indicates that the array status was written to the super.dat file? What are the other possible causes of this?
March 4, 200818 yr Author Do you have any drive letters assigned to the unraid from your PC? I have my itunes folder on my unraid with an allocated device letter assigned to it as well as another for scripted backup. Are you asking if I have a share from unRAID mapperd to one of my PC's? If so, no I don't. I periodically need to stop & restart the server in order for the user shares to be updated and I find that it never shuts down cleanly, on restart a parity rebuild is initiated. If I stop this parity rebuild & restart again the parity appears just fine again. Have you tried stopping that parity rebuild and restarting the array again? Mark. I haven't tried that yet, but that would make me doubt whether or not my data was actually protected. How could the parity actually be invalid and a restart cause it to then be valid? I realize it's just a bit or something that tells unRAID that the array is invalid and that it could be wrong at times, but if it returns an arbitrary value sometimes then can we ever be sure it's right when it says parity is valid?
March 4, 200818 yr Author I had the exact same problem. I posted the way I fixed it. http://lime-technology.com/forum/index.php?topic=1439.0 I hope this helps you. Phil I tried swapping the parity drive's cable between my onboard controller and my PCI SATA controller. I even replaced the cable just in case, but the problem persists. Thanks for the suggestion though.
March 4, 200818 yr I'll need to see the syslog as it shuts down & then reboots... To capture it during shutdown, via telnet session type this: tail -f /var/log/syslog Then click Stop on the web interface, then click Reboot. The last message should say something like, "sending DHCP_RELEASE...". Now select/copy/paste the text from the telnet window into a text file, call it 'syslog1.txt'. After system has rebooted, via console or telnet session type: cp /var/log/syslog /boot/syslog2.txt You can then find this file via the 'flash' share. Please post both log files.
March 5, 200818 yr Author I'm currently copying some data to that server. I'll take a look at that as soon as it's done. Thanks for taking a look at this Tom.
March 5, 200818 yr There is a line in syslog1.txt that explains why upon reboot a parity check is started: Mar 4 18:29:45 Tower1 kernel: [176886.401062] write_file: error 2 opening /boot/config/super.dat As system was being shut down, driver was not able to write 'super.dat' file in order to indicate 'clean shutdown'. So upon reboot, since 'clean shutdown' flag was not set, system started a parity check. 'error 2' means, 'no such file', which doesn't make sense... Also in the log you have a disk behaving badly. Appears to be a Samsung assigned to device (sdf) - it is reporting errors trying to read temperature from the drive. I wouldn't think the two things are related, but perhaps they are. Next thing to do is, from telnet session, type this: ls -lR /boot >/boot/flash.txt This will produce a directory listing of the entire flash into the file 'flash.txt' which will be created in the flash share. Please post.
March 5, 200818 yr You are getting the same sequence of errors for your Samsung HD753LJ that others are. See these threads: http://lime-technology.com/forum/index.php?topic=1499 http://lime-technology.com/forum/index.php?topic=1472 I think the answer may be to turn SMART on for the Samsung, at the bottom of your go script. Online, they are reporting a lot of compatibility problems with these drives and nVidia motherboard chipsets, and here with a Promise card. They also say they are very cool, very quiet, and very fast.
March 5, 200818 yr I also don't see how the Samsung problem could be related to the parity syncs, but did having to redo parity each reboot start about the time you installed the Samsung?
March 6, 200818 yr Author There is a line in syslog1.txt that explains why upon reboot a parity check is started: Mar 4 18:29:45 Tower1 kernel: [176886.401062] write_file: error 2 opening /boot/config/super.dat As system was being shut down, driver was not able to write 'super.dat' file in order to indicate 'clean shutdown'. So upon reboot, since 'clean shutdown' flag was not set, system started a parity check. 'error 2' means, 'no such file', which doesn't make sense... Also in the log you have a disk behaving badly. Appears to be a Samsung assigned to device (sdf) - it is reporting errors trying to read temperature from the drive. I wouldn't think the two things are related, but perhaps they are. Next thing to do is, from telnet session, type this: ls -lR /boot >/boot/flash.txt This will produce a directory listing of the entire flash into the file 'flash.txt' which will be created in the flash share. Please post. Here are the results: Tower1 login: root [Disconnect bypassed -- root login allowed.] Linux 2.6.22.5. root@Tower1:~# ls -lR /boot >/boot/flash.txt -bash: /boot/flash.txt: Input/output error root@Tower1:~# I also tried accessing the flash through Windows Explorer and it doesn't show any files. I tried to create a new folder on it and got this error: Unable to create the folder The request could not be performed because of an I/O device error I have the Flash share set to Export read/write so that's not the problem. I think maybe the flash drive is going bad.
March 6, 200818 yr The error message is telling you the flash drive is not writable. (actually, it cannot create the file... it could be the folder is not writable or not searchable) Could you try just typing ls -lR /boot You will need to cut and paste the resulting output from the telnet window (Hightlight text to copy, press right mouse button to copy to clipboard) The fact that the files are not visible from windows might be because you have windows set to not show "system files" It could be that the flash drive is no longer writable (read: going bad), but at least lets eliminate a folder or file permission issue first. Joe L.
March 6, 200818 yr Author Tower1 login: root [Disconnect bypassed -- root login allowed.] Linux 2.6.22.5. root@Tower1:~# ls -lR /boot /boot: total 0 root@Tower1:~# I have Windows set to show system files (I should've mentioned that). In fact the same computer can see the contents of my second unRAID tower's flash just fine. ...but did having to redo parity each reboot start about the time you installed the Samsung? No, I noticed it once or twice before I installed that drive.
March 6, 200818 yr Interesting... the ls command returns nothing... Please try the following three commands: ls -ld /boot and mount and ls -l /dev/disk/by-label Joe L.
March 6, 200818 yr Author Tower1 login: root [Disconnect bypassed -- root login allowed.] Linux 2.6.22.5. root@Tower1:~# ls -ld /boot drwxr-xr-x 3 root root 4096 Mar 4 21:35 /boot/ root@Tower1:~# mount /dev/sdg1 on /boot type vfat (rw) none on /sys/fs/fuse/connections type fusectl (rw) /dev/md9 on /mnt/disk9 type reiserfs (rw,noatime,nodiratime) /dev/md3 on /mnt/disk3 type reiserfs (rw,noatime,nodiratime) /dev/md4 on /mnt/disk4 type reiserfs (rw,noatime,nodiratime) /dev/md2 on /mnt/disk2 type reiserfs (rw,noatime,nodiratime) /dev/md6 on /mnt/disk6 type reiserfs (rw,noatime,nodiratime) /dev/md1 on /mnt/disk1 type reiserfs (rw,noatime,nodiratime) /dev/md7 on /mnt/disk7 type reiserfs (rw,noatime,nodiratime) /dev/md5 on /mnt/disk5 type reiserfs (rw,noatime,nodiratime) /dev/md8 on /mnt/disk8 type reiserfs (rw,noatime,nodiratime) shfs on /mnt/user type fuse (rw,nosuid,nodev) root@Tower1:~# ls -l /dev/disk/by-label /bin/ls: /dev/disk/by-label: No such file or directory root@Tower1:~#
March 6, 200818 yr OK... here is what I can see. Your ls -ld /boot command output looks fine. It is exactly the same as mine. The output of your "mount" command does not show that the flash drive is mounted on /boot. It should have an entry for it in addition to those for your disk and user shares similar to this: Oops... I missed it.... it was there in your mount command as /dev/sdg1 mounted on /boot root@Tower:/# mount /dev/sda1 on /boot type vfat (rw) none on /sys/fs/fuse/connections type fusectl (rw) /dev/md9 on /mnt/disk9 type reiserfs (rw,noatime,nodiratime) /dev/md5 on /mnt/disk5 type reiserfs (rw,noatime,nodiratime) /dev/md2 on /mnt/disk2 type reiserfs (rw,noatime,nodiratime) /dev/md4 on /mnt/disk4 type reiserfs (rw,noatime,nodiratime) /dev/md6 on /mnt/disk6 type reiserfs (rw,noatime,nodiratime) /dev/md8 on /mnt/disk8 type reiserfs (rw,noatime,nodiratime) /dev/md3 on /mnt/disk3 type reiserfs (rw,noatime,nodiratime) /dev/md7 on /mnt/disk7 type reiserfs (rw,noatime,nodiratime) /dev/md1 on /mnt/disk1 type reiserfs (rw,noatime,nodiratime) shfs on /mnt/user type fuse (rw,nosuid,nodev) Now, unRaid mounts the flash drive on the /boot folder by finding it by its volume label. On my system I get this output for the third command: root@Tower:/# ls -l /dev/disk/by-label total 0 lrwxrwxrwx 1 root root 10 Mar 5 15:11 UNRAID -> ../../sda1 When you typed the command to list the devices by their label it was empty. So... apparently, since unRaid could not locate your flash drive by its UNRAID label, it could not mount it on the /boot mount point and eventually could not write to the super.dat file that should be there. (that would be there if the "config" folder in the flash drive was there.) However... I'm really confused, since usually when this occurs the networking does not work since some other config files expected in /boot/config cannot be opened either. Can you confirm, by putting the USB drive in a windows box, that you have set the volume label to "UNRAID" ? Joe L.
March 6, 200818 yr Author I pulled the flash drive out and checked it in my desktop and the volume is named "UNRAID".
March 6, 200818 yr Author I tried to access the flash drive from Windows Explorer and I can now se all the files on it. Since someone mentioned the SMART temp problem (the drive never did show a temp in the web GUI) I decided to go ahead and copy smartctl to the root of the flash and then ran: /boot/smartctl -a -d ata -s on /dev/sda I now have a temp showing for that Smasung drive.
March 6, 200818 yr Apparently you plugged the flash drive back into your unRaid server and now it is identified and mounted. And you can see the files on it since it mounted properly... I'll bet the ls -l /dev/disk/by-label works too. All I can think of is an intermittent connection of the flash drive. Can you try a different USB port? Are you unplugging the USB flash drive after your server is booted? (you should NOT unplug it... it should be left in place all the time) If the connector you are plugging into has a cable connecting it to the motherboard, is the cable plugged into the MB securely? Perhaps Tom will have an idea with the extra clues you have supplied. Joe L.
Archived
This topic is now archived and is closed to further replies.