Janddy

Thanks, that makes sense to me. I wasn't sure between btrfs and zfs for the download disks, but I'm guessing the benefits of zfs won't really apply when you've got 2 disks and the data isn't that important such as downloads. And I think I'll probably do RAIDZ1 for the other pool so I have more space and room to grow. The files will be backed up externally anyway in case a drive fails.

I am about to rebuild my server, and I am looking for guidance on how to best utilise my drives and configure my pools. I can fit 20 drives in my case, and I have picked the most useful 20 I have. This is what I'm currently planning, but not set in stone. Main Array (xfs) 5 x WD Red 26TB (one of these as parity) 2 x Seagate Ironwolf 12TB 2 x Hitachi Deskstar 8TB 1 x Seagate Ironwolf 8TB 1 x Seagate Barracuda 4TB Appdata (btrfs) 2 x Samsung 970 Evo Plus 500GB in Raid 1 (mirror) The above is how my current server is configured, except without the WD Red drives, and the appdata pool is also used as cache for the main array. Since it is only 500GB though, the Samsung 970s often fill up when I download a lot in a short space of time. With the new build, I will have much more space and ports for drives, so I was thinking of adding a few other pools. Cache pool (btrfs/ZFS??) 2 x Intel D3-S4510 960GB SSD in RAID 0 or JBOD Having 2TB just for cache should stop the appdata drives from filling up with downloads. Other SSD Pool (ZFS??) 4 x Intel D3-S4510 480GB SSD 1 x Crucial MX300 500GB SSD My best idea so far with these drives is to use them as storage for data that would benefit from increased speed. I'm thinking maybe Photos for Immich, Nextcloud files, Paperless documents. Having said that, I currently don't have any speed complaints with Nextcloud or paperless (they are just simple documents after all), and I haven't deployed Immich yet so I'm not sure how much SSD speed will matter there. Currently my Nextcloud files are 311GB and my photos are 430GB. Since this would be important data, I would set it up with some sort of parity protected array or RAID config and maybe ZFS to help protect it. I do run a lot of containers (76) but the only other one I can think that might benefit from fast storage is frigate. Not sure if that's a better use case than the above. I also plan to do Plex transcoding in RAM. So how does this plan sound? Any obvious mistakes, or better file system/RAID choices? Do you have any other suggestions on what I should use the SSDs for, or a better way to configure all these drives? The drives that didn't make the cut were a 2TB Seagate Barracuda Green, a 1TB WD HDD, and 2 x Intel DC S3510 240GB SSDs. I figured adding a few faster SSDs would be more useful than 3TB of extra HDD space.

Ok will do, thanks!

Ok I have had another crash, and I think I worked out why it's happening. It's the same issue as my last post 😅 It seems my LiteLLM container is using too much memory. I started the LiteLLM container and watched the server RAM usage go from 52% up to 96% over about 10 minutes. I confirmed with htop that LiteLLM was using all that extra RAM. After that the whole server froze the same as before and I had to reset it with the restart button. I'll have to try and figure out if there's some kind of memory leak with the container or if it's the way I've set it up, or maybe it's normal and I just need way more RAM. As a fallback, is there a way to confine a container to only use a maximum amount of RAM? I've attached the syslog anyway if you're interested. syslog-10.1.1.10.log

Ok will do, thanks.

Hi, in the last few days I have started experiencing unraid unexpectedly freezing and being unable to interact with it. It has happened 3 times now I think - randomly - but usually while I'm logged in to it. When it happens I can't load the web UI, can't load the web UI of any docker applications, and DNS on my network devices stops working due to the pihole docker stopping. I can ping the server from my computer, but that's all. I noticed this issue start a few days ago after I had an issue while downloading lots of files, and the mover wasn't running frequently enough which caused my cache drive to fill up to 100%. My docker volume is a directory on the cache drive, so a few of the dockers crashed and had errors, which makes sense, but I'm not sure what's causing the instability with the operating system now. Would somebody be able to have a look at my diagnostics and try to help me figure out what's causing the problem? I had to press the restart button on the server to get it to reboot and get the diagnostics. Hopefully it didn't erase the info you need. Thankstower-diagnostics-20250806-1244.zip

I only use ones like ChatGPT, Claude, and Gemini though. I use OpenWebui to access them all from a single interface. I avoided the local ones like Llama etc. because my server isn't very powerful...I guess that didn't work haha.

Ok that makes sense. I didn't realise running big models used a lot of RAM - I thought it all happened in the cloud lol. I'll do some testing with different models and try to work out what the server can handle. Thanks for your help.

Oh ok. Yeah I have rebooted since the first post - 25 days ago. Probably something similar happened again since then. I've attached new diagnostics if you don't mind taking a look? tower-diagnostics-20250729-2019.zip

I'm still getting this error, but still haven't noticed any issues. Any advice?

Hi, Fix common problems is telling me I have an out of memory error, and to post here asking for help. This only started yesterday. I haven't noticed any disruption...yet. Any advice would be appreciated. Diagnostics attached. Thanks in advance. tower-diagnostics-20250601-2248.zip

Oh my goodness. It's all there. Such a relief! Thank you so much! I guess the mount configs must be saved on the flash drive or something and restarting just resets them to that config? Now I think I will just recreate the docker image and stop trying to balance it to solve that problem.

Oh, that sounds hopeful! I was scared to reboot in case it made things worse. And yes I think I'll take the simple option of rebooting if it will work just as well. Here goes...

I have been having issues with my docker image filling up. I deleted orphaned images and volumes etc. which seemed to help, but the docker settings page was still saying there was no available space and my server kept freezing. So I started trying to check for issues in the docker image. I ran the scrub on the docker settings page, then I ran btrfs check --readonly /mnt/user/system/docker/docker.img and then attempted to balance the docker image. This is where it gets bad. I tried to mount the docker image to run the balance and used the command mount -o loop /mnt/user/system/docker/docker.img /mnt This replaced everything in my /mnt directory, so it looks like all my data is gone. I know what I did was incredibly stupid and careless. I was rushing and overtired and not thinking. But I don't even know where to start with trying to recover from this. Is there anything I can do to try and get the data back? Thankfully I do have cloud backups of the most important stuff, but I would still hate to lose the rest of the data (mostly media), and starting again from the backup would not be fun. Thanks in advance. tower-diagnostics-20241221-0038.zip

I was on 6.9.2 so I just upgraded to 6.10.3 and tried adding the URLs again. The icons updated and then survived a force update and a server restart so it looks like that fixed it. Thanks for the help, I can't believe it was that easy in the end! I should have upgraded sooner.