May 17, 201115 yr I was in the process of writing to the array (copying files) when my workstation lost networking. I tried stopping the array but the disks cannot be unmounted. I followed the WIKI to manually kill the process that is hanging each disk (all disks there was the cache_dir process and the sleep). Then I issued the manual umount command for each disk. I then ran this command again fuser -mv /mnt/disk* /mnt/user/* to see if there were any more hung processes, but the command froze my terminal screen. I tried control-c to break out, but no luck. Only the unRAID web screen is responsive but it is just displaying Stopping..., unMenu is dead. I have IPMI and so can open another terminal session. In addition, I have two preclear process running. What can I do now to restart the array?
May 18, 201115 yr Author The preclear are in the Post-Read 75%. Would the disk be recognized as precleared if I powerdown the server? Would this cause any damage? The unRAID main page is reporting stopping, but I can no longer issue any commands on /mnt without freezing another tty session. I am completely stuck and may resort to just powering down the machine anyway I can and hope for the best.
May 18, 201115 yr Author Luckily the powerdown script worked and the array came back up. Everything looks fine except the one disk that failed the copy is not listed in the unMenu file browser. I can cd to it from the command line, but if I tried to copy files from Windows 7, everything freezes. Hopefully someone can give me some hints what to try.
May 18, 201115 yr Author I am getting a lot of these in the syslog when booting: May 18 00:29:27 Beanstalk emhttp: shcmd (25): set -o pipefail ; mount -t reiserfs -o noacl,nouser_xattr,noatime,nodiratime /dev/md5 /mnt/disk5 2>&1 | logger May 18 00:29:27 Beanstalk emhttp: shcmd (26): set -o pipefail ; mount -t reiserfs -o noacl,nouser_xattr,noatime,nodiratime /dev/md3 /mnt/disk3 2>&1 | logger Does means anything? Also attached is my syslog after a clean boot. Is there anything to indicate why writing to disk16 in the array would cause /mnt/disk16 to be unresponsive to even a simple ls command? syslog_mifronte_20110518.txt
May 30, 201115 yr I am also having trouble shutting down. I hit stop, and it is retrying forever. I logged in and tried to find what process is using a share with fuser (this usually works, and usually reports either cache_dirs or sabnzbd/sickbeard/couchpotato). This time, it just froze. cache_dirs, sabnzbd/sickbeard/couchpotato are all not running, but no matter what I try I cannot get out of the fuser process, and it's still retrying to unmount disk1. disk1 was exported as AFP/time machine. A backup was started to the disk, then stopped, and the disk ejected from my Mac. kill -9 won't work on the fuser processes. I closed the terminals that were running them, and they are still hanging around. I don't think fuser is preventing the shutdown, but I don't know what is. I tried running /sbin/powerdown and got a few lines of output about SMART overall-health self-assessment test resut: PASSED, and then nothing. The machine still responds to ping, I can still login as root over SSH, I can still access the unraid and unmenu web interfaces (but only by their IP address, not the server name). What else can I try, besides pulling the power?
May 30, 201115 yr I have this process in the output of ps -eO user -H: 4941 root D ? 00:00:00 /usr/sbin/cnid_dbd /mnt/user/Time Machine/.AppleDB 3 1 The file /mnt/user/Time Machine does not exist. Looks like the process is in uninteruptible sleep, and I have no way to kill it besides rebooting. But I can't reboot until I unmount the filesystem, which I can't do until I kill this process?
May 30, 201115 yr I have this process in the output of ps -eO user -H: 4941 root D ? 00:00:00 /usr/sbin/cnid_dbd /mnt/user/Time Machine/.AppleDB 3 1 The file /mnt/user/Time Machine does not exist. Looks like the process is in uninteruptible sleep, and I have no way to kill it besides rebooting. But I can't reboot until I unmount the filesystem, which I can't do until I kill this process? Sorry to say, "power switch"... If you cannot kill the process since it was written in a way that make it not interruptable, you are at its mercy. Joe L.
June 4, 201115 yr Sorry to say, "power switch"... If you cannot kill the process since it was written in a way that make it not interruptable, you are at its mercy. Joe L. I have exactly the same problem again. I made sure I shut down anything I was running on the machine (SABnzbd, CouchPotato, SickBeard) and tried to stop the array. It is now stuck in an infinite loop. It did "Stop SMB...Stop NFS...Stop AFP...Stop AVAHI..." which disconnected my shared drives. I tried fuser -c /mnt/disk1 again, which hung. The output of "ps -eO user -H" again shows: 27277 root S ? 00:00:00 /usr/sbin/cnid_dbd /mnt/user/TimeMachine/.AppleDB 3 1 "kill" did nothing. "kill -9" put the process into uninterruptible sleep. Now I'm screwed again. This folder does exist. Time Machine has been working correctly for the past few days. But it seems that when used with Time Machine, the .AppleDB folder is accessed in a way that is not closed properly when file sharing services are stopped? What is this cnid_dbd command? Why is it always still running even after all file sharing services have been stopped? Does this mean that I am going to have to perform a full parity check every time I reboot, if I ever want to use my unRAID array to store Time Machine backups? Cheers. UPDATE: http://cs.roosevelt.edu/eric/NetAtalk_on_Centos.html Seems to indicate that NetAtalk (which unRAID uses?) might not be compatible with Time Machine calls. Perhaps Time Machine makes a call that NetAtalk doesn't understand, which causes the process to hang and which subsequently prevents safe shutdown? Since unRAID allows shares to be exported over AFP (Time Machine), this seems to be a bug (or limitation) that unRAID should attempt to fix (or work-around)? Perhaps unRAID needs to use a patched version of NetAtalk? Here are some more links that may be relevant. http://www.sharedknowhow.com/2008/05/installing-netatalk-under-centos-5-with-leopard-support/ http://pastebin.com/f47499d34 Also, I have attached a copy of my syslog (while it is still attempting to unmount user shares). It has many lines about cnid, including one near the end that says: Jun 4 12:13:53 unraid cnid_dbd[30782]: main: not enough arguments I guess this may have occurred when I issued the "kill -9" command. syslog.zip
June 4, 201115 yr I'm backing TM without problem but I only run a few basic add-ons. I know others are backing up with TM as well without problems. IDK why this keeps happening : May 31 14:26:28 unraid afpd[26158]: login tailee (uid 1000, gid 100) AFP3.1 May 31 14:26:28 unraid afpd[26158]: Creating '/mnt/user/TimeMachine/Network Trash Folder' failed in /mnt/user/TimeMachine: Permission denied May 31 14:26:28 unraid afpd[26158]: Creating '/mnt/user/TimeMachine/Temporary Items' failed in /mnt/user/TimeMachine: Permission denied May 31 14:26:28 unraid afpd[26158]: Creating '/mnt/user/TimeMachine/.AppleDesktop' failed in /mnt/user/TimeMachine: Permission denied May 31 14:26:28 unraid afpd[26158]: Error opening /mnt/user/TimeMachine/.AppleDesktop/.volinfo: No such file or directory May 31 14:26:29 unraid afpd[26158]: Creating '/mnt/user/TimeMachine/Network Trash Folder' failed in /mnt/user/TimeMachine: Permission denied May 31 14:26:29 unraid afpd[26158]: Creating '/mnt/user/TimeMachine/Temporary Items' failed in /mnt/user/TimeMachine: Permission denied May 31 14:26:29 unraid afpd[26158]: Creating '/mnt/user/TimeMachine/.AppleDesktop' failed in /mnt/user/TimeMachine: Permission denied May 31 14:26:29 unraid afpd[26158]: Error opening /mnt/user/TimeMachine/.AppleDesktop/.volinfo: No such file or directory May 31 14:26:29 unraid afpd[26158]: getquota: special /mnt/user/TimeMachine fails I don't have anything like this in my server log.
June 4, 201115 yr After restarting, I see the following processes. root@unraid:~# ps -eO user -H | grep cnid 15021 root S pts/0 00:00:00 grep cnid 11722 root S ? 00:00:00 /usr/sbin/cnid_metad 12857 root S ? 00:00:00 /usr/sbin/cnid_dbd /mnt/user/Time Machine/.AppleDB 3 1 So it looks like cnid_dbd is a child process of cnid_metad. During the shutdown process, cnid_metad is stopped but it leaves it's child process dangling in a state that cannot be terminated?
June 4, 201115 yr It does not give me a problem. root@Rack:/mnt/user# ps -eO user -H | grep cnid 12699 root S pts/0 00:00:00 grep cnid 1822 root S ? 00:00:00 /usr/sbin/cnid_metad 5613 root S ? 00:00:20 /usr/sbin/cnid_dbd /mnt/user/TM-backups/.AppleDB 4 2 root@Rack:/mnt/user# It could be because of the permission errors.
Archived
This topic is now archived and is closed to further replies.