Jump to content

[SOLVED] User / User0 and missing files


Recommended Posts

Hi,

So I just upgraded to 5.0.6 this afternoon. I installed as per a new flash drive (and did a full format on the flash) and copied over my config dir and plugins and reinstalled basic packages (unmenu & preclear script).

Now the funny thing is that when I browse to one of my shares from windows, it only displays one sub folder when there should be 4. At first I thought all the files in the share were gone, but when I telnet into unraid and browse using midnight commander I can see all the folders as there should be on the path /mnt/user/SHARE-NAME/SUB-FOLDER. However, the path /mnt/user0/SHARE-NAME/SUB-FOLDER appears to be what I can see from my windows machine - that is to say this path has only one sub folder beneath the share name.

Can anyone tell me how to get access back to my files from windows.

I was going to try an unraid reboot, but for now I am running a pre-clear on a new but unassigned drive.

 

Thanks.

 

EDIT - I just looked again and now can see all files in MC under user and user0 but I still cannot access them from my windows machine.

Link to comment

I upgraded from version 5. Do I still need to do something with my permissions?

 

As for samba "root" user not working, is the option to set up a new user for accessing shares from windows? If so how to I go about this process? I really am learning as I go along with all of this...most of the time my server just sits there and works!!

Link to comment

UPDATE: it's not just windows machines that are affected I just tried accessing that share from a file browser app on iPad and it can only see one folder just like from windows. Also it only seems to be the one share that is messed up. Every other share seems to be behaving just as it was before!!

Link to comment

I think it was during the betas for v5 that root access to network shares was removed. I think the new security model was also introduced during the v5 betas. Were you upgrading from some beta v5 or was it released stable v5?

 

Is the share set for Public access? If so then the user shouldn't matter.

 

Go in to the webGUI and check the settings for that share.

Link to comment

I'm pretty sure it was stable. I tend not to really use the beta versions. Yes it's set for public access. All of my shares are. I have double checked this though and I also set it to private and switched it back to public just to be certain.

 

Is it possible that this is just a glitch that will fix itself with a server reboot? Should I try this once the preclear has finished doing its thing?

Link to comment

Looks like I'm going to have to wait for the pre-clear to finish before I can do any further testing, when I logged into web gui and hit stop array, it got stuck in a cycle where a message was scrolling along the screen saying something like "retrying unmount user shares" so there was a problem even stopping the array. Now I cannot access the web gui any more. Screen is still active running my preclear so when this is done, I'll shut down the server and reboot and then report back. Thanks for help so far.

Link to comment

Looks like I'm going to have to wait for the pre-clear to finish before I can do any further testing, when I logged into web gui and hit stop array, it got stuck in a cycle where a message was scrolling along the screen saying something like "retrying unmount user shares" so there was a problem even stopping the array. Now I cannot access the web gui any more. Screen is still active running my preclear so when this is done, I'll shut down the server and reboot and then report back. Thanks for help so far.

That's what happens if something has files or folders open on the array, including a console or telnet session's current directory.
Link to comment

I have found that if I have an active screen session then unRAID will not unmount the user shares. I always have to reconnect to screen and type "exit" then everything unmounts. Although in your case you don't want to do that because you are currently running a preclear.

Link to comment

I have found that if I have an active screen session then unRAID will not unmount the user shares. I always have to reconnect to screen and type "exit" then everything unmounts. Although in your case you don't want to do that because you are currently running a preclear.

 

Thanks - this is good to know, one less thing to worry about. I am 30% done on the final step of the pre-clear. Then I can reboot the server and resume the trouble shooting. Thanks again for all the help so far!

Link to comment

Ok so now I am really confused. The pre-clear is complete and my server is back up. A parity check started but for now I have just cancelled it. The share that I have been having problems with is called "Books" and it should have 5 subfolders. When I open it in MC it only has 4 subfolders. But there is a new share that I had not noticed before, I didn't create it, which is called "books" (with a lower case b). This has only one subfolder which is empty as it currently should be. When I browse my network I can see a share called "Books" with an upper case B but the contents of this folder appear to be the contents of the new share that is listed in MC and webgui with a lower case b. I'm scared to delete the new lower case b share as when I changed the properties (to set it to export-no) the other share changed its properties to match so it seems like they are linked.

 

Any ideas about this?

 

I'm thinking I should create a completely new share - copies all the files over and then delete the randomly behaving shares with the upper and lower case Books/books name and then rename my new share. Will that work or is that just ignoring a potential problem?

Link to comment

You can probably get this fixed using MC. The thing to remember is that any folder at the top level of cache or any array disk is automatically considered a user share. Just use MC to move things from one disk to another so you get everything that is under the books share into the Books share. It is very important to use the disk paths for this move and not the user paths; i.e., /mnt/diskX/books/subfolder to /mnt/diskY/Books/subfolder. If you use the user paths you run the risk of losing data.

 

Once you have everything consolidated into the Books share, the webGUI should let you delete the books share.

Link to comment

I've never encountered the behaviour you are describing. Linux is case sensitive so books and Books are not the same thing. Windows can get confused by this as you have seen. unRAID will not let you delete a user share that is not empty.

 

Look on your flash drive and see if there is a share cfg file for each of these at config/shares/books.cfg and config/shares/Books.cfg

Link to comment

That ties in with what I was trying to explain earlier, I didn't ever create a share called "books", only one called "Books" so I don't know where "books" came from and if the .cfg file being missing means that I didn't configure the share then that would be correct as I never added this share in the first place.

 

Could it be possible that the "books" share was created by an app called Mylar (kind of like sickbeard or couchpotato but for books) which I recently added. Maybe a lower/upper case typo on my part when entering a file path could have resulted in this odd kind of duplicated share behaviour?

 

If this is the case I am maybe best transferring my files to a new share e.g. BooksNEW and then deleting the badly behaving share(s) when empty and then renaming BooksNEW to Books. Do you think this would work to tidy things back up again?

Link to comment

Creating a new share is unnecessarily complex. Use MC to browse to each disk in turn and find the rogue books folder and rename it to Books. You are probably correct in your supposition about mylar. Check its settings and verify that it is configured to save to Books, or else you will just have this issue again the next time it runs.

Link to comment

Yes, if you setup a program to create new content and used books instead of Books then you'd get exactly what you are seeing. Samba and Windows can't distinguish between books and Books so they get all confused. But, the server is fine and the data on the server is fine.

 

So, do as already suggested. On the server using Midnight Commander (mc) and navigate to each disk then either re-name the rogue books directory to Books or copy the data from the rogue books directory to the existing Books directory. Once you've taken care of the data you can just delete any remaining rogue books directories. Once you've cleaned it up then stop and start the array or reboot and the books share will be gone.

Link to comment

The problem is now fixed - thanks for all the help!! In the end I had to move all files (not very many) out of "books" and back into "Books" and then I deleted "books" and its all back to normal as I had some issues renaming a folder in MC, it seemed to be moving the folder rather than renaming it! Thanks again for the help.

Link to comment
I had some issues renaming a folder in MC, it seemed to be moving the folder rather than renaming it! Thanks again for the help.

Move and rename are the same thing in this particular instance. You are trying to move the contents from a folder named books into a folder named Books and delete the original books folder. It's semantics, but it really does the same thing.
Link to comment

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...