Hi everyone and thanks in advance.

 

I have some old TV show files that cause issues with PLEX because they are read-only it seems. They are always in the 'recently added' section and cannot be marked as watched/unwatched etc.

 

I've tried deleting them with Dolphin, Windows and from terminal (as per screenshot attached) with no luck. I've tried again after running Docker Safe New Permissions... nope. I can see that it's not permissions, it's a read-only problem but I can't find how to tackle it. 

 

With my limited knowledge, I considered moving all other files off the disk, making the disk unmountable and formatting then moving everything else back but that seems like shooting sparrows with cannons. Any better suggestions?

41 minutes ago, Husker_N7242C said:

Any better suggestions?

Post your diagnostics.

The chances are that you have some sort of file system corruption (which will be repairable using the correct tools or the format of the affected disk) on one of the disks in the share.   The diagnostics should allow the problem disk to be identified.

2 hours ago, Husker_N7242C said:

making the disk unmountable and formatting

Not clear what you have in mind here. Any disk in the array must be formatted by unRAID. You can't format it any other way and maintain parity. And formatting a disk that's already in the array is easy, no need to "make it unmountable", whatever that means.

Hi Johnnie, here is the diagnostics

nas-diagnostics-20180310-0955.zip

8 hours ago, trurl said:

Not clear what you have in mind here. Any disk in the array must be formatted by unRAID. You can't format it any other way and maintain parity. And formatting a disk that's already in the array is easy, no need to "make it unmountable", whatever that means.

Hi Trurl, Thanks for your help. Making it unmountable can be done by stopping the array and changing the disk file system, for example from ReiserFS to XFS. The disk will then show as unmountable and then you can format. I believe that this maintains parity. I'd rather learn to use some file system tools to fix this if possible

10 hours ago, itimpi said:

The chances are that you have some sort of file system corruption (which will be repairable using the correct tools or the format of the affected disk) on one of the disks in the share.   The diagnostics should allow the problem disk to be identified.

Thanks itimpi, I hope the diagnostics shed some light. My biggest worry with UNRAID is a fear of the unknown, which for me is the worry that data becomes quietly corrupt without me knowing. Some 'getting started' with file system tools would be great. I'm hesitant to implement things like Nextcloud properly before learning a bit more about maintaining the health of the file systems.

4 minutes ago, Husker_N7242C said:

Making it unmountable can be done by stopping the array and changing the disk file system,

 

When the file system is changed, a disk becomes unmountable because the "new" file system isn't initialized yet. The next step is to format the disk and make the file system ready for use. While doing a format disk, parity will be updated accordingly. As a matter of fact parity is always updated in real-time whenever something is written to the array.

 

There's filesystem corruption on disk2:
 

Mar  9 00:48:11 NAS kernel: REISERFS error (device md2): vs-5150 search_by_key: invalid format found in block 450590991. Fsck?
Mar  9 00:48:11 NAS kernel: REISERFS (device md2): Remounting filesystem read-only

 

run reiserfsck.

Edited March 9, 20188 yr by johnnie.black

Also: syslog is completely filled with USB disconnect errors and you need to adjust cache floor:

 

Mar  9 01:15:18 NAS kernel: usb 3-4: Device not responding to setup address.
Mar  9 01:15:18 NAS kernel: usb 3-4: Device not responding to setup address.
Mar  9 01:15:18 NAS kernel: usb 3-4: device not accepting address 13, error -71
Mar  9 01:15:18 NAS kernel: usb usb3-port4: unable to enumerate USB device
Mar  9 01:15:25 NAS shfs/user: share cache full
Mar  9 01:15:25 NAS shfs/user: share cache full

 

3 minutes ago, Husker_N7242C said:

Hi Trurl, Thanks for your help. Making it unmountable can be done by stopping the array and changing the disk file system, for example from ReiserFS to XFS. The disk will then show as unmountable and then you can format. I believe that this maintains parity. I'd rather learn to use some file system tools to fix this if possible

It's true that if you change the filesystem, then it is unmountable as the new filesystem until it is formatted. I believe it would still be mountable as the old filesystem.

 

Of course, I'm completely aware of how to change the filesystem, having converted all my disks long ago and helped others do the same. I've never heard anyone describe it as "making it unmountable", but I guess you could look at it that way.

9 minutes ago, johnnie.black said:

adjust cache floor

For the OP, this is in Global Share Settings. You should set Minimum Free for cache and for any user share to be larger than the largest file you expect to write.

1 hour ago, johnnie.black said:

There's filesystem corruption on disk2:
 

Mar  9 00:48:11 NAS kernel: REISERFS error (device md2): vs-5150 search_by_key: invalid format found in block 450590991. Fsck?
Mar  9 00:48:11 NAS kernel: REISERFS (device md2): Remounting filesystem read-only

 

run reiserfsck.

Thanks Johnnie again. You helped me last time as well and again really quickly and straight to the point. I really appreciate it. 

 

Is this corruption something that is avoidable through maintainance?

 

Thanks trurl also for your tips. 

 

I'm not sure what's the deal with USB. Could this be the keyboard and mouse switching between VM and GUI when the VM is started and shutdown? I haven't noticed any real issues with USB 

 

I don't quite understand what the cache floor issue is. I recently expanded the cache to 2 drives. I'll do some reading

 

6 hours ago, Husker_N7242C said:

Is this corruption something that is avoidable through maintainance?

Filesystem corruption doesn't usually happen (or it shouldn't) without a reason, like unclean shutdown, memory issues, etc, you might also want to look into converting to xfs as reiserfs is no longer recommended.

 

6 hours ago, Husker_N7242C said:

I'm not sure what's the deal with USB. Could this be the keyboard and mouse switching between VM and GUI when the VM is started and shutdown? I haven't noticed any real issues with USB 

It doesn't appear to be the keyboard or mouse, and if don't notice any issues it's not serious other than filling your syslog, you could try disconnecting all USB devices one at at time to try and find which one it is.

 

6 hours ago, Husker_N7242C said:

I don't quite understand what the cache floor issue is.

Check trurl's reply above.

9 hours ago, Husker_N7242C said:

I don't quite understand what the cache floor issue is

On second though ignore this one, this:

 

share cache full

Only means you're hitting that share's minimum free space setting, i.e., it will start writing directly to the array, I confused that one with:

 

cache disk full

This one means you're hitting the cache floor, which is by default 2GB and would be very close to filling up your cache disk.

 

2 hours ago, johnnie.black said:

This one means you're hitting the cache floor, which is by default 2GB and would be very close to filling up your cache disk.

 

If that is the default, then it is too low for many purposes.

 

To the OP,

Minimum Free tells unRAID when it should choose a different disk to begin writing a file. In general, it has no way to know how large a file will become, so if a disk hasn't hit minimum free yet it can choose the disk to begin writing the file and if the file becomes too large the write will fail.

 

As mentioned, cache and each user share has a setting for this, which should be set larger than the largest file you expect to write.

2 hours ago, johnnie.black said:

On second though ignore this one, this:

 

share cache full

Only means you're hitting that share's minimum free space setting, i.e., it will start writing directly to the array, I confused that one with:

 

cache disk full

This one means you're hitting the cache floor, which is by default 2GB and would be very close to filling up your cache disk.

 

Johnnie, again, you're awesome.

 

Filesystem... I was already converting disks to XFS but skipped disk2 as I had trouble getting the data off (suprise suprise). 3 out of 13 disks to go!

 

Cache floor, i think I've got the cache free space set to about 20GB which is about where it is sitting. I have appdata set to 'preferred' so that apps (mainly plex) just use the array for metadata once the cache is full. I plan to add a third cache disk when time allows, just in raid 0 aswell seen as most of the cache is backed-up weekly at worse (the VMs everyday).

 

Corruption... I only recently replaced my UPS and had a power outage recently (caused by my wife pointing a hose at a powerpoint out the back lol). This did cause issues and to restore backups of the vms and appdata. Not totally suprising that there is corruption. Thanks for finding which disk! I'll do some reading and see if I can clean it up.

 

USB.... I haven't got around to getting my USB3 PCIe cards setup properly with IMMOU groups, etc so I just have a single keyboard/mouse set which is allocated to the VM. When the VM is off they seamlessly revert to controlling UNRAID GUI (I've been booting in GUI mode for now because of the issues I had with the VMs after the power 'failure'. Does the disconnection error mean that the USB device is being seen as disconnecting itself unintentionally or could it be cause by simply shutting down the VM then using the keyboard and mouse in the GUI?

29 minutes ago, trurl said:

If that is the default, then it is too low for many purposes.

Yeah, note the that share minimum free space takes precedence, still would also like to see a higher default, like at least 10GB