Hey!

So for a while ago, maybe 3-4 months - i added an cache drive in my array.
Some of the new files first lands on the cache disc and then every wednesday it gets moved to the array - this has been working fine for some months now, but recently i noticed that everytime it does this, my whole array just freezez - i can't see any dockers and everything is super slow, the UI even craches sometimes, but as soon as it's done moving, everything moves on as usual.

I don't really know where to start searching for the errors, i don't get any parity errors or something like that..

I have provided my latest diagnostic.

 

Thanks!

slave-diagnostics-20230823-2035.zip

You have too much memory.

 

But seriously I too have 128GB and encounter similar issues and I believe the only fix is to adjust the write caching to ram. I have been searching this issue for weeks and lots of people have similar issues and the only resolution seems to be the following:

 

Download the tips and tweaks plugin and adjust down the vm.dirty_ratio and vm.dirty_background_ratio values. Or set those values yourself from the command line.

 

I need to do some more indepth testing but I currently have mine set to 4 and 2. I haven't fully tested if it fixes my specific symptoms when running the mover since I have undertaken other migitations to resolve them but I do know I am seeing similar I/O waits with the new values too.

 

If you want to find previous topics search for 'io wait' or 'dirty_ratio' but here is the one in particular that led me to this fix:

 

Edited August 24, 20232 yr by sjrahn

23 hours ago, sjrahn said:

You have too much memory.

 

But seriously I too have 128GB and encounter similar issues and I believe the only fix is to adjust the write caching to ram. I have been searching this issue for weeks and lots of people have similar issues and the only resolution seems to be the following:

 

Download the tips and tweaks plugin and adjust down the vm.dirty_ratio and vm.dirty_background_ratio values. Or set those values yourself from the command line.

 

I need to do some more indepth testing but I currently have mine set to 4 and 2. I haven't fully tested if it fixes my specific symptoms when running the mover since I have undertaken other migitations to resolve them but I do know I am seeing similar I/O waits with the new values too.

 

If you want to find previous topics search for 'io wait' or 'dirty_ratio' but here is the one in particular that led me to this fix:

 

 

Wow, too much memory! Didn't think that should make any errors, only solutions! hehe.. 
I have read more about I/O waits and just like you say, it seems to be the problem.

The only thing confusing me, is that it was working fine before, and i haven't changed anything since then... i was first starting to think some disk in my array was causing this.

 

I have changed the dirty_ratio and i hope this solves the issue, thank you alot for the response!

Edited August 24, 20232 yr by DjJoakim

On 8/24/2023 at 10:06 PM, DjJoakim said:

 

Wow, too much memory! Didn't think that should make any errors, only solutions! hehe.. 
I have read more about I/O waits and just like you say, it seems to be the problem.

The only thing confusing me, is that it was working fine before, and i haven't changed anything since then... i was first starting to think some disk in my array was causing this.

 

I have changed the dirty_ratio and i hope this solves the issue, thank you alot for the response!

 

Soo, just for an update... This didn't fix the issue. Every time my mover starts it kills 60% of my dockers and freezes unraid