Hi all,

 

Im running in to an issue i was hoping i could get some help with. 

 

Specs and usecase:

Unraid 7.0.0, intel 12500, 2TB M2 cache, 3x 3.5" EXOS CMR HDD, 2.5Gbit network. 

xfs on array drives, no parity. Appdata, vm's etc is on the SSD. HDD's only contains media.

My media is highest quality available, mostly 4K remux. I have around 10 plex users, almost all with high bandwidth internet connections and modern devices. Not unusual to have a few users streaming 4K remuxes all around 150Mbit. This normaly works great.

 

When the mover runs and is writing to a disk that Plex is also reading media from, the stream will buffer. It seems like the mover "totaly consumes/overrides" all other disk activity. I have tried some fixes from the forums/reddit, none seem to really work. 

 

Tested with Priority for mover process and Priority for disk I/O but they dont seem to make a noticable difference, still unable to stream from plex during moving.

 

Found something about setting vm.dirty_ratio to 1, doesnt help for me.

 

When downloading a few large files over smb share to windows pc in my network, im having no trouble streaming over plex. bandwidth/disk io seems to be shared somewhat evenly between everything. Im not sure if this is because all processes are reading from the disk, not writing. Writing would go to the cache drive, so no issues there.

 

Moving large amounts of files between disks with "unbalanced" plugin causes the same issue as the mover. almost totaly consuming all disk io/usage.

 

After some googling, this seems to be an issue for years. I can schedule the mover to run at a convenient time but i have users streaming at different times so i would like to avoid situations where users are effected by this, i want them to always have a good experience using plex.

 

There must be some way to just set the mover to go at like 50MB/s or low priority or something? I dont care if the mover needs to run a few hours longer, i just want it to be super low priority. 

 

If i need to provide more information let me know.

Thanks in advance!

 

 

 

 

Edited February 7, 20251 yr by freakyfreek

Schedule mover for idle time. 

 

Default mover schedule is once per day in the middle of the night. 

Is there a way to actualy tell unraid to schedule the move during an idle moment or are you telling me to manualy schedule it at a time that is likely to be idle? 

34 minutes ago, freakyfreek said:

to manualy schedule it at a time that is likely to be idle? 

This.  The default is early hours of the morning.

Well i know i can schedule it at a different moment, i addressed this in my post. This doesnt resolve the issue.

I agree with OP.  I've been trying to use the new feature of mover in 7.0 to empty a drive but it seriously makes everything else lag.

I would still like some help with this. Thanks!

If that's an issue for your server, why not use the mover during off hours, like when you are sleeping?

I already addressed this in my post and again in the comments, why are you suggesting it again? 

 

It's simple. Plex won't work during moving. Plex will be used any hour of the day. There are no off hours. 

 

I have it scheduled at a time where plex usage is least likely, but i would still like an actual solution, not a workaround. 

 

Thanks! 

30 minutes ago, freakyfreek said:

I have it scheduled at a time where plex usage is least likely, but i would still like an actual solution, not a workaround. 

As far as I know there is no way to dynamically schedule when mover should run.

Writing to the array (such as when moving), while also trying to access the array for reading, will likely involve some competition for array disks. With seeking involved in that competition, which is the slowest thing disks do. This is really just a hardware limitation.

 

Possibly if you could guarantee the files were being written to a disk that wasn't being read it might be better.

 

If you are trying to use turbo write, it might help to turn that off since it is going to try to read all disks to update parity.

 

https://docs.unraid.net/unraid-os/manual/storage-management/#array-write-modes

 

If you really need speed, a striped zfs pool instead of the Unraid parity array would be a possibility.

Im not using parity. So its raw HDD speed. 

 

The issue is the mover proces doesnt consider other processes. When going trough SMB, or any other "file usage" really, this issue doesnt happen. It seems really strange to me that 1 process is allowed to hog a disk like that. I have never encountered this in a different scenario. Why is it not "shared bandwith / IO" like trough SMB? If i copy 10 files from a single drive, they will all go at like 20MB/s, not 1 at 199.90MB/s and the other 9 at 0.01MB/s. What is the process/command that is actualy used when the mover is running? Maybe i can look for a general resolution with the process/command being used on linux.

Have you considered bypassing your cache pool and just writing data directly to the array? You're not going to find a simple solution to this issue if you don't compromise either by running mover at a time of least plex use or bypassing the cache and downloading/unpacking to a single disk.

Edited March 25, 20251 yr by MowMdown
spelling