October 14, 201411 yr I run unraid 6 beta 10a and am currently transferring my disks from ReiserFS to XFS. Because of that I frequently have to stop the array. On every occasion so far, I am on the 8th disk now, I experience that my server becomes unaccessible from the Windows browser. I can't see Tower in Windows Explorer. I have tried booting in safe mode, with or without Xen, its always the same. The ip-adres is still alive though. I can still access the server in putty or IPMIview. I then reboot the server with powerdown -r and all is back to normal. It is just very annoying that I have to do this every time I finish the transfer of a disk to XFS and want to start reformatting another disk. What causes can there be that this is happening? Something else that I found is that booting with Xen in safe mode still loads SABNZBD and other plugins.
October 14, 201411 yr There are a couple of things it can be worth trying. Clear the browser history in IE. There is a known problem with long cookies causing problems. I believe that Tom is planning a change to emhttp to help alleviate this issue. Use an alternative browser (.g. Chrome).
October 14, 201411 yr Author There are a couple of things it can be worth trying. Clear the browser history in IE. There is a known problem with long cookies causing problems. I believe that Tom is planning a change to emhttp to help alleviate this issue. Use an alternative browser (.g. Chrome). Thanks for the speedy answer. This is only concerning the Internet Explorer. The Tower also disappears in Windows (disk)explorer and network environment when stopping the array in my case.
October 14, 201411 yr The Tower also disappears in Windows (disk)explorer and network environment when stopping the array in my case. That is normal ehavior. When you stop the array you are closing down the network services on the unRAID server. You should then get access to the 'flash' share as that is visible even with the array stopped.
October 14, 201411 yr Author The Tower also disappears in Windows (disk)explorer and network environment when stopping the array in my case. That is normal ehavior. When you stop the array you are closing down the network services on the unRAID server. You should then get access to the 'flash' share as that is visible even with the array stopped. Yeah, well the entire server disappears and consequently so does flash. I used to edit my go file and upgrade to a new version by accessing flash.
October 14, 201411 yr The Tower also disappears in Windows (disk)explorer and network environment when stopping the array in my case. That is normal ehavior. When you stop the array you are closing down the network services on the unRAID server. You should then get access to the 'flash' share as that is visible even with the array stopped. Yeah, well the entire server disappears and consequently so does flash. I used to edit my go file and upgrade to a new version by accessing flash. What happens if you try and put the path to the flash in the File Explorer address bar (\\serveraddress\flash). I have seen Windows do funny things in the past.
October 14, 201411 yr Author Never ever had to do that. Also strange that the server name disappears from Windows network environment. It happens on two different laptops.
October 15, 201411 yr Once again though, if you Stop the array, you will not be able to access any shares (flash drive is a share \\tower\flash). You should still be able to get to the webgui from web browsers to start array. I finished migration to xfs and it was not that bad, while array was started, copied files from one disk to a spare. Stopped array, click the reiserfs disk I wanted to convert and changed to xfs and started array and checked the box to format it. Shares are available at that point and I copy files from the spare back to the disk and repeat.
October 15, 201411 yr Author I understand that I lose access to my shares when I stop the array. But my server (Tower) should not disappear from Windows network environment and flash also shouldn't disappear. But both do. I also can't access the servers WEBGUI from either Firefox or Internet Explorer on both Windows laptops. Just rebooted with a stock go file in SAFE mode and started copying to a fresh XFS formatted disk. See how it works out this time
October 15, 201411 yr I think this functions as it was 'designed' and implemented. When you stop the array, unraid stops all samba-related processes which includes nmbd which I think deals with providing the name of the server for windows NetBIOS and network neighborhood functionality. Unless you have a DNS/bind server also running with the name to IP translation of your server, windows will not find the server when the array is stopped. LimeTech will have to weigh in on this as they'd know for sure.
October 15, 201411 yr Author I guess samba is closed before it starts stopping the array? So during the stopping of the array and unmounting disk sequence there is no samba? I saw several times many retry umount messages in a syslog. Perhaps a plugin or unmenu package makes this happen? Like I said before I booted this time with a stock go file in SAFE mode, so without unmenu, packages and without plugins. Hope it will stop the array properly on the next occasion (over 24+ hours .
October 17, 201411 yr Author Still the same trouble as described earlier. Even with a stock go file and safe boot. I started a parity check last evening which finished this morning without error. Then I made a backup of the USB key. I deleted all files except super.dat, pro.key and some cfg files, I did delete disk.cfg though, from the usb stick and copied over the unraid6 beta 10a files from the zip file, like when creating a new usb key. So basically I have a fresh unraid6 beta 10a install now, without unmenu, packages and plugins. The server booted without starting the array. I manually started the array, which also wrote to disk.cfg. I hope this helps to solve some of the trouble I experienced. The syslog looks clean now.
October 17, 201411 yr I saw several times many retry umount messages in a syslog. If you see a never ending cycle of this it indicates the "Stop Array" task never completed and hense samba never gets restarted with Flash only. This will also eventually cause emHTTP to stop responding in my experience.
October 17, 201411 yr When you lose access to the server from Windows, can you get to it by using the !P address rather than the server name in File Explorer? If so that would indicate that the [problem is related to resolving NetBIOS names within your network.
October 18, 201411 yr Author I saw several times many retry umount messages in a syslog. If you see a never ending cycle of this it indicates the "Stop Array" task never completed and hense samba never gets restarted with Flash only. This will also eventually cause emHTTP to stop responding in my experience. The BIG question is: What causes these endlenss umount retry unmounts shares messages?
October 18, 201411 yr An add on or plug in that is running. It doesn't have to be accessing the share at that point either.
October 18, 201411 yr Author Like I said earlier, I use a clean install of V6 beta 10a, without unmenu, packages, plugins whatsoever. The same happened this morning after I copyied another disk to XFS. I stopped the array and an endless sequence of "unmouting shares, retry unmounting shares" appeared. I hope it isn't hardware related. I also found out that powerdown isn't doing anything. Do I need a newer version for V6b10a?
October 18, 201411 yr Are there any other PC's or devices on the network that may be keeping the share alive? Like a media player, laptop? Also, can you throw up a syslog that has this event in it?
October 18, 201411 yr Author Are there any other PC's or devices on the network that may be keeping the share alive? Like a media player, laptop? Also, can you throw up a syslog that has this event in it? I try that next time. What is the easiest way to get a syslog without unmenu installed? Would restarting my router help to see if other devices are keeping a share busy? All network activity is disrupted then for several seconds.
October 18, 201411 yr OK, been trying to follow this thread but a few things aren't clear to me: 1 - so when you stop the array, nothing is accessible over the network, is that correct? 2 - When you start the array again after stopping it to reformat a disk, it doesn't show up on the network via SMB anymore, is that correct? 3 - rebooting resolves this issue? 4 - Please post a syslog.
October 18, 201411 yr Author OK, been trying to follow this thread but a few things aren't clear to me: 1 - so when you stop the array, nothing is accessible over the network, is that correct? Yes, but this is probably because stopping the array doesn't complete, because of numerous retry unmount shares messages. 2 - When you start the array again after stopping it to reformat a disk, it doesn't show up on the network via SMB anymore, is that correct? I have to reboot the server by powerdown -r to do this. When the server is rebooted it is back on the network and I can set another filesysten on a disk and reformat. 3 - rebooting resolves this issue? Yes, see point 2 4 - Please post a syslog. Is in the attachment, current running a parity at which many rcu_sched self-detected stall on CPU messages show up in the syslog. syslog-2014-10-18.zip
October 18, 201411 yr That syslog is a mess! Crash after crash, 50% of it by length is the output of crashes. Have you tried with 5.0.5?
October 19, 201411 yr Author That syslog is a mess! Crash after crash, 50% of it by length is the output of crashes. Have you tried with 5.0.5? What mess do you mean: The "rcu_sched self-detected stall on CPU" messages? They only appear short after starting a parity check and stop after a couple of hours . I see this mentioned in in more threads, like here: http://lime-technology.com/forum/index.php?topic=34973.0 . I also saw them in V5 betas and stopped appearing in V5 final upto V5.05. Going back to V5.05 is not an option as I have half of the array converted to XFS. The parity check finished this morning without errors with a total runtime of 13 hours and an average speed of 83 MB/s. After it finished I could stop the array without problem, set a new disk to XFS, format it and start moving data to it from another disk in Midnight Commander. There is no occurrence of a parity check in the syslog because the parity check was initiated automatically after an unclean shutdown of the server.
Archived
This topic is now archived and is closed to further replies.