I run a MythTV server that NFS mounts an unRAID share (video2) as one of the locations it uses to store recordings.  This has worked fine for quite some time.  A few months ago, I upgraded to unRAID 5.0.5, at which time I decided to enable a cache drive, figuring that would work well with MythTV since it writes a large amount of data as it records an HD program.  I created a cache drive using a spare 80GB Seagate 7200rpm drive that is otherwise too small for a spot in the unRAID cluster.  Later, as I was doing a one-time bulk load of some photos and/or music data to the unRAID array, I found that the cache drive was filling up, so I went into the unRAID console and changed the mover schedule from once a day at 3:40am to hourly, simply by changing the 3 to an asterisk.  As a result, the mover was running at 40 minutes past the hour, every hour.  Seemed to be working fine. 

 

As time passed, I began to notice that *some* MythTV hour-long recordings were being truncated.  At first, I wasn't sure what was causing this, but then I realized it was only happening on the recordings being stored directly to the unRAID server.  When I viewed the recording, it would get to about 40 minutes into an hour-long show and then hang, eventually crashing the playback client in MythTV.  At this point the light bulb went off and I realized that the mover script also runs at 40 minutes past the hour, right about the same time that these recordings are being truncated.

 

This makes me wonder if there is an issue with the mover script that is allowing it to start moving a file that is currently open and being written to.  Could there be something about the way my NFS mount is set up, or the way MythTV writes to the file that causes this?  Or could there be a flaw in the mover script that allows it to start moving a file that is open for update via NFS mount? 

 

Attached is a screenshot of my share configuration.

 

Larry

Anyone?  This could be a potentially serious flaw in the mover script.

 

 

Sent from my iPad using Tapatalk

Sorry. I saw this, but know NADA about NFS.

I've not seen that behavior on SMB or AFP.

 

(Could it be something with MythTV? (I don't use that, either).

Sounds as if MythTV is closing the file frequently, so when the mover runs its sees the file as ready to be moved. But that's just a theory.)

 

Dale, it occurred to me that if MythTV closes the file periodically, this could happen.  Seems unlikely, though.  I thought it could be a quirk with NFS. I'll see what I can find out over in those forums...

 

 

Sent from my iPad using Tapatalk