Hi, I'm looking for some advice please.

 

I'm new to Unraid, and pretty new to Linux as well, being a long-time windows user.

 

I currently have an Unraid setup, with only 1 NVMe disk in the array, with a Home Assistant VM, and a number of docker containers running on it (Usenet/Arr stack, Plex, Frigate for security cameras, to name a few). I don't currently have any plans to create a large storage array, as I really only want to use Unraid for the simplicity of running docker containers and VMs. I don't really care about storing Movies and TV Shows once I have watched them.

 

I have noticed an issue with SABnzbd, where my download speed will drop significantly during downloads, and also direct unpacks taking a very long time. I believe this may be because the write speed is limited on disks in the array, however, I am not 100% sure about that, as I don't have a parity drive, and I'm not sure if that is the only contributing factor. This is not happening on my main Windows 11 machine. SAB is saying my download and complete folder speeds are only around 90MB/s, but DiskSpeed docker is showing proper speeds of around 2300MB/s. I have gigabit internet, and downloads will start at around 109MB/s, but drop to around 20MB/s.

 

I have a second, unassigned, 1TB HDD as well, and I was thinking of moving the NVMe drive to a cache pool (I believe it should perform better then) and putting the HDD on the array. I will then run all of my docker containers and VM on the cache.

 

So, I am trying to figure out the simplest way of doing that, without losing data, and moving my containers and VM over to the cache. I have tried searching, but I can't really find anything that matches my situation exactly.

 

I read about the unbalanced plugin, and have copied all of the data to the other disk, so if I were to clean my NVMe now, create a new config, assign the HDD as disk 1, the NVMe as cache, and then use unbalance again to copy the data to the NVMe again, how do I run the containers and VM from the new location?

 

Is that even necessary? Could I just create a new config, and move the NVMe to a cache pool with the data still on it?

 

Any help would be greatly appreciated, thanks!

You should post your system's diagnostics zip file in your next post in this thread to get more informed feedback.  It is always a good idea to post this if your question might involve us seeing how you have things set up or to look at recent logs.
 

Here is my diagnostics zip

server-diagnostics-20240129-1150.zip

SSDs on the array cannot be trimmed, so it would be better using it with a pool, also recommend using exclusive shares to bypass FUSE, than can help a lot with performance.

Am I right in thinking that the share needs to be in a pool before I can set it to exclusive?

Correct, pool only.

Ok, cool, thank you.

 

So what is the simplest way of moving my ssd, while keeping all of my shares, dockers and vms intact, to a ache pool?

 

Is it as simple as setting a new config, preserving all assignments, and moving it to a cache pool? And then replacing disk 1 in the array with my other HDD? Or is there more involved?

20 minutes ago, Jonathanpbk said:

Is it as simple as setting a new config, preserving all assignments, and moving it to a cache pool? And then replacing disk 1 in the array with my other HDD?

Yes, you can leave just disk2 in the array, you only need a single array device for now.

Sorry for the delayed response. Thanks for your help, I now have my NVMe moved to a cache pool, and my HDD on the array, with all shares etc working as they should.

 

Unfortunately, I am still having performance issues with SABnzbd, so I am really scratching my head. Without direct unpack, my download speeds aren't tanking any more, but unpacking is still taking a long time, even just on the 10GB test download. Even though my download/complete folder speeds are now reporting a much more reasonable (though still not full speed) ~740MBs.

If you are using exclusive shares not much more that can be done on the Unraid side, could be a container specific issue, or an NVMe device issue.

Yeah, I figured as much. Might try asking in the hotio discord, to see if I'm missing something with their container.

Thanks for your help, anyway!

On 1/30/2024 at 11:05 AM, Jonathanpbk said:

Yeah, I figured as much. Might try asking in the hotio discord, to see if I'm missing something with their container.

Thanks for your help, anyway!

 

Did you find a solution? I'm kind of seeing the same situation as you.