Unraid Running Slow

[Chandler]

My server has been running slow for a while now and I am tired of dealing with it acting this way. Dockers will be slow to start/respond, I'll get upstream failures in the logs when trying to load pages on the GUI, server will occasionally hang on reboots, the list can go on but I'm just wondering if anything stands out in my diagnostics. 

tower-diagnostics-20200713-1436.zip

[JorgeB]

Nothing obvious, I would recommend updating to latest stable, if still issues try booting in safe mode with everything off and then if OK start enabling dockers/plugins/services one at a time.

[trurl]

Your appdata, domains, and system shares have files on the array, so your docker and VM performance will be affected by the slower writes (and slower disks) in the array, and dockers and VMs will not allow array disks to spin down since they will always have open files on those shares.

 

Whether that is enough to cause all your symptoms I don't know, but I would suggest getting those shares all on cache and configured to stay all on cache.

 

[Chandler]

52 minutes ago, johnnie.black said:

Nothing obvious, I would recommend updating to latest stable, if still issues try booting in safe mode with everything off and then if OK start enabling dockers/plugins/services one at a time.

Alright I've just updated and tested a few things.

 

The main thing I am using to test is Plex. For some reason, only on iOS connections, everything I play buffers every 1-5 seconds. I thought it was Plex so I troubleshot that first with the verbose logging to check my transcode speed but when I'm on wifi it is a direct play and there is no transcoding but it was buffering which was odd. My network can easily handle this. I switched to cellular and it was transcoding 1080p 10,000kbps bitrate to SD and struggling according to Plex verbose logging which said I was at around 0.3 when I want that number to be above 1. What is odd is that my CPU usage was only around 25% and RAM around 25% as well. So that is when I thought it must be something acting up with the Unraid server among the other problems I am having overall.

 

Anyways, after updating and rebooting my above problem was still present so I tried booting in safe mode. I attempted the above with Plex and it worked flawlessly on wifi and when switching to cellular, the verbose logging said I was transcoding at a speed of 13.5 which sounds much better. 

 

So safe mode boots without plugins, correct? Would my next step be to figure out what plugin may have been causing this issue? How do I enable the plugins one at a time? I'm not even sure if it did boot in safe mode because my syslinux config is already reset to boot back into normal Unraid and not safe mode (unless it does that automatically) and all my dockers started on their own and it appears all plugins are still working. 

17 minutes ago, trurl said:

Your appdata, domains, and system shares have files on the array, so your docker and VM performance will be affected by the slower writes (and slower disks) in the array, and dockers and VMs will not allow array disks to spin down since they will always have open files on those shares.

 

Whether that is enough to cause all your symptoms I don't know, but I would suggest getting those shares all on cache and configured to stay all on cache.

 

Appdata is set to prefer cache so I'm not sure why there are files being put on the array when I have 2TB of cache.. I'll set it to only use cache as well as the other shares you mentioned. 

Edited July 13, 2020 by Chandler

[trurl]

Just now, Chandler said:

Appdata is set to prefer cache so I'm not sure why there are files being put on the array when I have 2TB of cache.. I'll set it to only use cache as well as the other shared you mentioned. 

You can see which disks are being used by these shares by going to the User Shares page and clicking Compute... for the share.

 

Simply setting these to cache-only will not get their files off the array. You must set them to cache-prefer and run mover. But since mover can't move open files, you must also disable docker and VM services in Settings before running mover.

 

And mover won't move duplicates, so it is possible you might have to do some manual cleanup after moving.

 

If you need further help with this post new diagnostics before re-enabling dockers and VMs so we can see what remains to be done.

 

 

[Chandler]

32 minutes ago, trurl said:

You can see which disks are being used by these shares by going to the User Shares page and clicking Compute... for the share.

 

Simply setting these to cache-only will not get their files off the array. You must set them to cache-prefer and run mover. But since mover can't move open files, you must also disable docker and VM services in Settings before running mover.

 

And mover won't move duplicates, so it is possible you might have to do some manual cleanup after moving.

 

If you need further help with this post new diagnostics before re-enabling dockers and VMs so we can see what remains to be done.

 

 

Okay, I set domains and system to cache-prefer and started the mover after turning off docker and vms. My appdata was already set to cache-prefer and has been since I started using Unraid. I find it odd that some files exist on the array.. 

[trurl]

Prefer shares will overflow to the array if cache gets to Minimum Free in Global Share Settings. 

[Chandler]

Just now, trurl said:

Prefer shares will overflow to the array if cache gets to Minimum Free in Global Share Settings. 

Ah, I will have to check what I have that set to then. Thanks. Currently while watching the mover logs, it is saying some files can't be moved because it already exists. I look at its location on the array and its location on the cache but it only exists in one of them. The folder it needs to go in is on both but the file itself only exists once. Any ideas on this one before I start deleting things? 

[trurl]

13 minutes ago, Chandler said:

Any ideas

 

1 hour ago, trurl said:

mover won't move duplicates

Will have to look at the details 

[Chandler]

7 minutes ago, trurl said:

 

Will have to look at the details 

What details? 

 

Yeah, I understand it fails when attempting to move a duplicate. The issue is that I cannot find the duplicate. For example the mover is saying it is moving /mnt/cache/folder/fileA to the array but it already exists. I looked through /mnt/disk#/folder/ for fileA on each disk but it was not on any of them. From what I could tell it only exists on the cache. 

[trurl]

What do you get from the command line with this? (substitute real folder / filename as appropriate)

ls -lah /mnt/user0/folder/fileA