I'm putting together a new server and would like some recommendations on the storage setup. I've got a new Xeon w-1370, motherboard and 64GB of ram.
For storage though I was hoping to reuse parts I already had. Those include 4 10TB HDDs, 2 1.8TB Intel DC SSDs, and a 1TB pci-e gen4 M.2 SSD. At first I was thinking 2 mirror vdevs for the 4 spinning drives in one pool for mass storage and another mirror vdev with the SSDs for a faster pool to put any VM's in. But then I was looking into special allocation and thinking maybe I should try that instead with the SSDs. It would definitely be nice to not have to worry about 2 separate pools. I'm not 100% sure how that works, hence this post, but I figure I'd need to set special_small_blocks to actually really take advantage of the 1.8tb. I'm not sure how much storage just metadata would take up. Is there a way to estimate that?
The fast 1TB M.2 drive seems overkill for the OS drive of SCALE, but I already have it. I could use it for L2ARC or SLOG, but from most of what I've read I don't think that would be very useful? So I'm not sure what to do with that. Maybe use some of it as the transcode directory for the plex container.
The main duty of the server will be for fileshare, plex, and openhab. But it will also host a few dev VMs for my webapp projects (Hopefully LXC or LXD containers at some point). And a handful of other containers as well. The main purpose of putting together a new setup though is to be able to transcode 4k content in plex. From what I've read here passing through the iGPU to a container should be working. As well as to get familiar with TrueNAS SCALE. I've got FreeNAS Mini's deployed at several clients sites as backup storage and there are definitely a few that could take advantage of having a linux host instead of FreeBSD.
For storage though I was hoping to reuse parts I already had. Those include 4 10TB HDDs, 2 1.8TB Intel DC SSDs, and a 1TB pci-e gen4 M.2 SSD. At first I was thinking 2 mirror vdevs for the 4 spinning drives in one pool for mass storage and another mirror vdev with the SSDs for a faster pool to put any VM's in. But then I was looking into special allocation and thinking maybe I should try that instead with the SSDs. It would definitely be nice to not have to worry about 2 separate pools. I'm not 100% sure how that works, hence this post, but I figure I'd need to set special_small_blocks to actually really take advantage of the 1.8tb. I'm not sure how much storage just metadata would take up. Is there a way to estimate that?
The fast 1TB M.2 drive seems overkill for the OS drive of SCALE, but I already have it. I could use it for L2ARC or SLOG, but from most of what I've read I don't think that would be very useful? So I'm not sure what to do with that. Maybe use some of it as the transcode directory for the plex container.
The main duty of the server will be for fileshare, plex, and openhab. But it will also host a few dev VMs for my webapp projects (Hopefully LXC or LXD containers at some point). And a handful of other containers as well. The main purpose of putting together a new setup though is to be able to transcode 4k content in plex. From what I've read here passing through the iGPU to a container should be working. As well as to get familiar with TrueNAS SCALE. I've got FreeNAS Mini's deployed at several clients sites as backup storage and there are definitely a few that could take advantage of having a linux host instead of FreeBSD.