Kjetil Posted December 15, 2019 Author Share Posted December 15, 2019 (edited) Ok, I can install more drives if neccessary. I have normal hard drives available, or should I use SSD ? How many drives and how big drives are recommended ? The server has 4 or 5 spaces available for hard drives (its a 36 disc server, now using 30 or 31 drives). But not sure if these slots can be used and assigned for cache ? The cache drive now is not part of these slots (its an SSD inside connected by cable, and not using the slots / docks like the other drives). Edited December 15, 2019 by Kjetil Quote Link to comment
Kjetil Posted December 15, 2019 Author Share Posted December 15, 2019 Not sure if I can use more than 30 discs for the Unraid presently ? I think it was the max number of drives before ? Quote Link to comment
Kjetil Posted December 15, 2019 Author Share Posted December 15, 2019 28 disc + 2 disc for parity Quote Link to comment
trurl Posted December 15, 2019 Share Posted December 15, 2019 If you decide to add one more cache drive it ideally would be another SSD close to the same size as the one you already have. The total capacity of the pool would be equal to the size of the smaller of the SSDs. Other possible configurations are possible, but that is simplest and gives redundancy using default settings. If you think you might like to have a pool of more than a single cache drive, then that will restrict you to reformatting cache as btrfs. That is what many of us have for cache since it is the only choice for multiple cache pool. You can do that and add the additional SSD later. But that is getting ahead of ourselves. We need to get everything off your cache first. Quote Link to comment
trurl Posted December 15, 2019 Share Posted December 15, 2019 3 minutes ago, Kjetil said: Not sure if I can use more than 30 discs for the Unraid presently ? I think it was the max number of drives before ? Personally, I would recommend not even considering adding any more disks than you currently have. If anything you should try to get to fewer larger disks. I always recommend fewer larger disks instead of more smaller disks. Larger disks are more cost effective and perform better due to increased density. Every disk you have is a point of failure so the fewer the better. Quote Link to comment
Kjetil Posted December 15, 2019 Author Share Posted December 15, 2019 Ok, I am hoping 10TB or larger gets more affordable. But not sure if they will within the next year or so.... So what's the next step now ? Quote Link to comment
trurl Posted December 15, 2019 Share Posted December 15, 2019 From looking at your latest diagnostics it seems you have been working on some of these things yourself, so for example, your appdata is all on cache now. What we need to do next is get everything moved off cache so it can be reformatted. Just to make sure we don't miss anything, go to Main - Cache Devices and click on the folder icon at the far right then post a screenshot. Quote Link to comment
Kjetil Posted December 15, 2019 Author Share Posted December 15, 2019 (edited) I have only followed the instructions here on the forum, except for this I have done no other work. Here is the Cache devices screenshot. Edited December 15, 2019 by Kjetil Quote Link to comment
Kjetil Posted December 15, 2019 Author Share Posted December 15, 2019 Could the missing Plex info be on the cache ? Quote Link to comment
trurl Posted December 15, 2019 Share Posted December 15, 2019 11 minutes ago, Kjetil said: Here is the Cache devices screenshot. OK that agrees with what I had seen from looking at all of your shares cfg in diagnostics. Let me address your other question before continuing. Quote Link to comment
trurl Posted December 15, 2019 Share Posted December 15, 2019 2 minutes ago, Kjetil said: Could the missing Plex info be on the cache ? Nothing on cache is "missing". Cache is part of the user shares, and your appdata share is on cache (where it ultimately belongs). Probably what is wrong with your plex is a corrupt database, which is in the appdata, but not missing in any way. As mentioned you might have to start over with that. We will get there. Quote Link to comment
trurl Posted December 15, 2019 Share Posted December 15, 2019 The next thing you need to do is set those 3 shares that appeared in your latest screenshot (those that have files on cache) to cache-yes and run mover so we can get them moved off cache. After mover finishes post another screenshot just like that one so we can make sure nothing is still on cache before we reformat. Quote Link to comment
Kjetil Posted December 15, 2019 Author Share Posted December 15, 2019 Ok, understand. I am prepared to setup Plex again. Quote Link to comment
Kjetil Posted December 15, 2019 Author Share Posted December 15, 2019 Ok, so if I understand correct I go to "Shares" then choose each share and set "Use cache (for new files/directories):" to Yes ? And then how do I start the mover ? Quote Link to comment
trurl Posted December 15, 2019 Share Posted December 15, 2019 1 minute ago, Kjetil said: Ok, so if I understand correct I go to "Shares" then choose each share and set "Use cache (for new files/directories):" to Yes ? And then how do I start the mover ? Yes. Go to Main - Array Operation and click Move Now. Quote Link to comment
Kjetil Posted December 15, 2019 Author Share Posted December 15, 2019 Ok, mover is working. Not sure how long this 42GB will take to move... Quote Link to comment
Kjetil Posted December 16, 2019 Author Share Posted December 16, 2019 Cache move is done. Here is a screenshot of the cache. And a diagnostic. federalreserve-diagnostics-20191216-0941.zip Quote Link to comment
Kjetil Posted December 16, 2019 Author Share Posted December 16, 2019 Its ready for the next step Quote Link to comment
trurl Posted December 16, 2019 Share Posted December 16, 2019 OK. The whole point of course was to get everything moved off cache so it could be reformatted, so we are ready for that now. You have 2 choices, XFS or btrfs. If you think you might want to eventually add another cache disk then btrfs is the only choice. That is the only way to get redundancy for cache. Whether or not redundancy is important for cache is another question. Cache is where your system share and appdata will live. appdata share will contain all the "working" files for your dockers, such as your plex database. system share will contain docker image, which is easily recreated so no need to backup. If you ever have VMs system share will also contain libvirt image. There is a plugin that will automatically backup appdata and libvirt to the array. So, redundancy in cache for these is less important. If you decide to also cache some of your user shares, then those cached writes will have no redundancy unless you have another cache disk. But those cached writes will be moved to the array anyway. If you think you might eventually want to add another cache disk you have to choose btrfs for the current cache disk, or else you would have to repeat this whole thing later so you could reformat cache. Personally, I have 2x275GB SSDs as btrfs raid1 cache pool, but to some extent I just did this as a learning exercise. I could probably do just fine with a single cache disk and the backup plugin. And I don't cache user share writes since most of my writes are scheduled backups and queued downloads, so I am not waiting on those writes to complete anyway. By writing directly to the array they are already protected. Some people keep other things on cache just for performance or to avoid spinning up array disks. I have a copy of some of my music and photos in cache-only shares so they can be frequently accessed by other devices on my network. But since those are only a copy redundancy isn't needed. In addition to those and my appdata and system shares, I also use SSD cache as temporary storage for DVR since there is some benefit to the performance gain when trying to record and playback at the same time. Plex has DVR capability but it requires some hardware to get the signal to record. So, your choice, XFS or btrfs. btrfs is arguably the more flexible choice, and the default choice for cache on new installs. Quote Link to comment
Kjetil Posted December 16, 2019 Author Share Posted December 16, 2019 Sure, I have no knowledge of cache, so if btrfs is the recommended, then lets go with that. Can you see from the diagnostics (or some statistics) if the drives spin up often (or too often) ? I dont write that often to the server/array. If I write to cache first, can it be set up to automatically move - or do I manually have to move from cache to array ? Do I need to physically remove the cache drive to format it ? Or can we do it while its connected inside the server ? Quote Link to comment
trurl Posted December 16, 2019 Share Posted December 16, 2019 2 minutes ago, Kjetil said: Sure, I have no knowledge of cache, so if btrfs is the recommended, then lets go with that. Can you see from the diagnostics (or some statistics) if the drives spin up often (or too often) ? There is nothing in diagnostics specifically to track spin up/down. You might be able to get some information by carefully analyzing the syslog but I am not sure everything related to that is logged. 4 minutes ago, Kjetil said: I dont write that often to the server/array. If I write to cache first, can it be set up to automatically move - or do I manually have to move from cache to array ? The default schedule for Mover is daily in the middle of the night. That schedule is at Settings - Scheduler. Mover is intended for idle time. Some people try to cache too much and then expect Mover to keep up. It can't regardless of the schedule or even running it manually. There is no way to move from fast SSD cache to slower HDD parity array as fast as you can fill cache, and trying to write and move at the same time just causes competition for the disks. Better to not cache if you have a lot of data to write. Quote Link to comment
trurl Posted December 16, 2019 Share Posted December 16, 2019 12 minutes ago, Kjetil said: Do I need to physically remove the cache drive to format it ? Or can we do it while its connected inside the server ? You MUST format it while it is in the server and assigned as cache. Instructions to follow. Quote Link to comment
trurl Posted December 16, 2019 Share Posted December 16, 2019 I haven't needed to format anything in a while and I am not going to try it now just to refresh my memory. I may not describe it in exacting detail, but it is fairly simple. Before it will allow you to format anything, you must stop the array first. Go to Main - Array Operation and Stop. Then, go to Main - Cache Devices and click on the disk to get to its Settings page. Change the File system type to btrfs. Then go to Main - Array Operation and post a screenshot just to refresh my memory. Ask questions if this is unclear or there is something missing in my description. Quote Link to comment
Kjetil Posted December 16, 2019 Author Share Posted December 16, 2019 btrfs or btrfs encrypted ? Quote Link to comment
Kjetil Posted December 16, 2019 Author Share Posted December 16, 2019 brtfs, then choose apply ? Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.