Bizquick

This is the chipset on the motherboard. Marvell SE9172: 4 x SATA3 6.0 Gb/s, support RAID 0, 1, 5, 10 and a link to my mainboard if that helps. I dont want to start using it and find out later. https://www.asrockrack.com/general/productdetail.asp?Model=C246 WS#Specifications I'm planning on a redoing my system to use a ZFS Data array instead. And if I use that controller for that I wouldn't want to risk seeing drives falling offline. I would end up with issues with my zpool etc... I just figured that this has been a long time known issue that somebody must have figured out how to improve driver support by now.

Is Marvel controllers still bad to use in unraid with array drives? I only ask because my ASRock WS246 Board has 4 extra Sata ports using a Marvel 9172 controller. I was trying to trim down not using my HBA controller. But I could stick with using it because I'm using onboard GPU from the processor. But It would be nice to not have to use one of my PCIe slots just to hook up all the drives I would use. If its still an issue kind of surprised I would have thought something would have improved by now.

Man nice Motherboard Choice for a AMD proc. Just kind of sucks they didnt give you more SATA ports. I would have expected better from a ASRock Server Rack board. You should be ok to add that kind of adapter card. I think others have used the same one. One question I have if this is going to be for a Media Server. Are you adding a Nvida GPU card for transcoding? I know recently they just got plex to with with ATI cards in Linux. But I'm sure its no where nearly as good as Intel QuickSync. Me personally I've been building on much older hardware spec's and using Intel CPU's with GPU. But AMD can be really good too. It just depends on what your building. From the sounds of things it sounded like you are looking to run low power as well. I think you might want to rethink the setup if your not planning to use a GPU card. For me I would look at a older Intel 10 to 13th gen with similar core count like a i5 something. I wouldn't really look into 14th gen because then you get into those new weird cores. I'm not sure anyone has like developed the OS to take any benefit from those yet. The preformace cores and stuff sounds good in theory. But everyone mostly just builds apps and dockers to the OS layer and depend on the OS to handle the new hardware changes. You might try your idea on a build. And then later on see how much better the ATI Graphics gets better for Transcoding. I would expect to wait almost 1 year before it gets better. But again if adding a GPU then it doesn't matter.

If use this plugin to backup my appdata. Can I use the plugin to restore my appdata to another unraid server? I'm looking for a easy way I can have a backup server ready to go with out rebuilding everything separately.

Wow seams like a lot of power just to be the front end server to run some dockers and VM's but I guess it depends on the VM's. From what you said you have your Synology NAS as your storage array. If your keeping that for your media storage. You should be okay with your setup. There is many ways you can do your connections to connect your data to your dockers or VM's The simple way is just creating SMB shares with Unassigned devices. Years ago when I was getting my feet wet with unraid. I needed to transition my setup from FreeNAS. And all my Plex Shares I just mounted it with a SMB shares till I figured out how I wanted to setup my Data array in unraid and copy my data. Its wasn't the best way I wouldn't recommend it on 1Gig nics if your going to stream 4K video. Worked fine for me on most cases. But at that time I don't think 4k video was even a thing. So I would say it depends on your I/O Speed and NIC speed for that kind of setup. Anyway. For what your spec'ed out I would do what what your planning put the 2 NVME's in the array and make one a parity. You can put your dockers and VM's all in your array because it sounds like your data is going to be in the synology NAS. you will want some sort of backup in case your parity was to fail you in the event the other NVME fails. You could find a script to backup the array to a SMB mounted share or write one. If you dont want to script anything and your not using to much memory for hosted VM's you could also use one of the many backup dockers to back up your array to your Data synology box. this would be one way of setting things up. It would work fine till you could build something to have 1 box do all of it. If I had this Idea your doing I would run like this for a while till I could build 1 nice box running unraid. And when done find some extra data drive I could stick in the Synology box and setup a script to back up the unraid arrays and turn off and on the synololgy box. many ways to do stuff like that though. I think this is the only way I could see using a Mini with some NVME's and unraid. And memory that's a good amount of memory unless your VM's need a large amount. if this was me and I just needed this for a Media front end like your talking about. I could do this with 32gigs easy. 16 gigs would be tight. But I might be able to pull that all if I didnt have VM's. Oh and I would use ZFS format on any of the array drives. it will just slow it down. With parity you will be in good shape. Good luck though. I might try to build something similar with some of my older hardware to get an Idea how that might work out. I just would have to sub the Synology with something else in my old hardware bins I got some Qnaps the company retired this year. I might use that in my test case. Oh my mini would be much older though only a Dell 7060 with 32gig ram. But I can make something similar.

Well that must be what is going on then because I can only see 8 drive bays and the 2 NVME's on the PCI-E Card. I would have no Idea how I could change the back plane to use Some sort of HBA connectors. I guess that would be my next attempt. I almost want to grab the 12 bay one and take it home and try that one instead. Only small problem is I don't have a small form Video card I can use to get into the Bios. None of these models have a external display port so I have to add a Video card to just get into bios to change the boot order. Also the 12bay the Processor is just a 4c with Hyper threading. So I would have to change the processors around just to get that. But If I find its worth it and all bays work. I might consider it. Thanks for checking that over and letting me know. Now I need to learn where to see what your talking about in the log to see that so I dont have to ask so many questions.

Sorry I was going to do a new thread but I think that could be a bad Idea. Here is the Diag. neonas4-diagnostics-20231101-1630.zip

Ok I'll do that and start a new thread. I tried reloading the server with TrueNAS Scales and same issue. I might try Core Version and see if BSD OS can see the drives. But I'm suspecting not. I have a feeling its most likely using a Marvel controller and it most likely will never work. Which is disappointing but if thats the case. This makes me a little miffed. I like this free piece of crap I got. Just dont like Qnap OS.

Well I dont think I'm going to get much help with this. But I thought I would give it a shot. I got this large peice of poop Qnap server. 24 drive bay. anyway it had 48gig of ESS ram and a Xeon 2136 CPU 6 core with Hyper Threads. anyway I add a video card so I could get to the bios to change the boot order. But I cant get any Linux OS's to see more than 8 Drive bays. Any suggestions?

I would hope so its the LSI SAS3816 chipset. Also there was no way I could even configure a raid array all drives were seen as just drives. I didn't have to do any pass thru crap or attempt to make an array. if even wanted the raid functions restored so I could make a raid 0 or a raid 1. I would have to figure out how I could cross flash it to LSI retail mode. I think too many bad users tried to flash their old cards to IT mode and Dell got a bunch of dead parts. So it looks like they are trying to lock that down. If you want HBA mode you buy the HBA mode controller. If you want IR mode you get the Raid mode controller. They sell this HBA controller for Azure MS servers apparently. we use it for Storage Space Direct on a few small MS clusters. I'm getting really good Speeds on it. I got 4 drives on 1 channel. and the other 2 drives on B channel. I got 215Megs avg when I did a PreClear on a 12TB drive. I usually get 160 to 175megs on my SAS3008-16i card. And that was putting putting 3 drives on port1 and 2 drives on Port 4 and parity on the Motherboard. But ZFS kills my speeds though. I have to get my one ZFS drive out of my main array. I like having it for making a place to clone my snap shots to. But Not worth it right now. I'll come up with another Idea. Over all I'm shocked to see the speeds above 200 when it runs for more than 10 plus hours.

So I got a interesting new piece of hardware I didn't have to buy. I got new new Dell HBA355 Adapter. I got all the firmware updated on it 2 months ago. About all I could do on that. Anyway uses SFF-8654 Connectors. So I had to get these break out cables. https://www.amazon.com/dp/B0B5CRTG52?psc=1&ref=ppx_yo2ov_dt_b_product_details Bad thing is I don't have a board that I can do PCIE 4.0 or a CPU for my server to do that. I just recently got switched over to a Xeon E-2226g and Running a ASRock Rack Server board 64 gigs ECC too. Anyway using this Controller I think it down stepped to PCIE 3.0. But any idea what performance I should expect with it on my spin drives? And or should I put my Sata SSD's on it too from my cache array? I'm just trying to get the best I/O I can expect. I was very shocked I couldn't setup any sort of raid array at all with this controller. Not like I want to do that. I want IT mode anyway. But I found it really weird that Dell locked it down to be just a HBA. Watch the days of just getting a Dell controller and re-flashing it could be going away.

I can answer some of this. First off your setup sounds like a good Idea to start with. But I have some suggestions. if you want drives to power off or trying to manage your power usage. Forget ZFS. Think of ZFS like this it software raid and software raid all the drives are part of an array and always running. Unraid's array is not a software raid array. It just a Bunch of Disks and turns on each drive that access's the data. For me so far after testing a few diffrent things. I have actually found XFS in the main array the best format for almost all situations. But depending on your cache drive setup you might want to make 1 of the drives a ZFS format. keep in mind the preformace for that 1 drive will actually suck pretty bad. At least it does for me. I could have something configured wrong but I'm pretty sure that ZFS just sucks in the main array because its really just a raid 0 single ZFS array on one drive. Anyway back to your cache drive. For me I've learned it really depends on what your using the cache drive for. For me I'm doing a media server and I like having redundancy. I use 2 small 512Gig NVME's for my AppData dockers. And then a secondary Cache using 2 Sata SSD's those are for my downloads and unpacking of Video files I get from NZB's etc. (Something your most likely not doing so ignore having a secondary cache pool. Most people use torrents and seed boxes now days I just do things odd and harder.) I would say this if you need your apps up and running pretty quick you should have 2 NVME's in a mirror. And for size It all depends on what the NVME is doing. If you hosting VM's on the NVME I most likely would mirror with 2 NVME's. But if its just dockers like me. You might not need more than 512 to 1 TB. And if you can handle a outage to replace your NVME. Then run with 1 and make it ZFS. And have at least 1 of your drives in the array formated for ZFS. It means that drive will always be on. But what you could do is you could follow Space Invader's video for setting up ZFS snapshots and make in backup the snapshots to that one ZFS disk in your array. That way you could recover your appdata pretty fast and easy if your NVME fails. and depending on the snap shot schedule you might not loose much or notice anything lost. But as for your cache drive to me I found if I'm just using it for dockers. Sata SSD's were plenty fast enough for me. But VM's though NVME way better. If it was me and I was limited on some of the funds. and I could afford 1 NVME or 2 Sata SSD's to use for cache and no VM's were involved. I would do 2 Sata SSD's in a Zfs Mirror array. But again that's only if I was doing dockers. I think once you get your feet wet and start with what your thinking. You will figure out what is best for you though in the end. That's the really cool thing about unraid so much you can do and easy to start with a little and go up. Oh one more thing though Memory ZFS will use a little bit more memory so if you were thinking of running with only like 16gigs. double that and don't get to mixed up in the ECC or non-ECC. Its not going to make a huge difference in most cases. I don't care who disagrees with me on that. But most of us are running home server setups. Unless your going to buy server grade parts just don't get mixed up in all that. If you guys want to have that argument go back to the A-holes at TrueNAS and live in those forums. Just last month I actually started to change all my gear to some Server grade parts. And I ended up spending 3 times what I would have spent. And that's even using ebay to buy some used gear. UDIMM ECC way to expensive to use for desktop processors or low end intel Xeon's. for 64gigs UDIMM PC3200 DDR4 ECC ram me around 220. I could have used desktop non ECC for 100 to 125 max. And that was used ECC too. But if I was running higher end Server Grade Intel Xeon's like a dell server or something. I could have used RDIMM's And ECC on those are way cheaper and many options. Anyway I'm way off topic. Your idea sounds good to start with just make sure your memory is like 32Gigs if you use ZFS at all.

Network transfers have allways been tricky for me to find the answers. But one thing i know I found my issues were mostly related to the source I/O. if your reading or writing to spin disc your I/O is your factor. If your reading and writing all to SSD's in both cases. Then I'm not sure what the issue is outside of some sort of network driver. But I have noticed what your talking about lately. example for me I'm 10 gig Networking too. And my SMB writes to and drom my cache array and my PC's NVME over SMB. Its dropped some in speed too. looks like I'll need to do some of the same tests you have been doing.

I'm still learning quite a bit. I've been playing with rebuilding my main box and using a secondary Data ZFS array. And just setting up my main array with just a single SSD drive formatted in ZFS. I've come to this understand that the product can be so custom fit for almost any design. ZFS has great benifts but for small scale setups I would only recommend using it for the cache arrays and maybe 1 drive the the main array. I had to go back and rethink about things and look at the product name and the intial intention of UNRAID. And I think the main focus on the product unraid. It was never intended to be a raid array. The creators of unraid designed the systems main array so you could loose some drives and still not loose all your data in your main array. They also designed it so you could set shares up to only write to certain drives. And then certain drives will not be used writing all the time and your I/O could be better in some cases depending on the design. You could save the spin life on some drives in some cases. But if your only going to use 3 spin drives for your main array. I wouldn't waste much time into ZFS thinking unless its for the cache drive or drives. I mean I might format one of the drives in the main array for zfs and zfs for the cache drives. and that's only so I could could follow Space Invader's recent video series for auto snap shots and ZFS backups etc.. But to be honest depending how your using the cache drive or drives you could do many ways to backup and recover. And the great thing is you can redo and change your setup latter. You might find that you want to experiment with VM's or lots of dockers etc.. Over all this product can be very customized to fit almost any idea. And you could make it with a little bit hardware or a lot. And there any many here that can offer ideas to fix issues.

Is this what I need to do if I followed SpaceInvader's old videos on setting up like my own BitWarden server etc... I'm just getting that message now in Fix Common problems and not sure what steps I'm going to need to take to switch from MacVlan to IPVLan.

Thanks for all that. you know out of everything not sure where. But in the end I found the only reason it was not working. was because I had to turn of the sharing at the share in unraid. The SMB extra settings I kept trying in were not applying. I spent so long doing many other things before trying that. They should say if you use the extra setting to do that in the Manual. It would have made so much more sense. Or better yet add allowing vfs object in the global smb.conf But I'm going to guess Limetech has way more things to fix before going back to SMB config stuff. They are most likely happy with how it works now and don't have time to look at global config issues.

Looks like a good option if not using like server boards. Something maybe I would use. I my self not having many server parts or able to have many unless I get a good deal. But lately with some of the parts I have been using I can only use 1 NVME Drive. and the other NVME has to be a Sata m.2. So I would say it depends if you want to pass a VM with a NVME or not. If you use one of these you need to make sure you double check what your mother board will allow and what you loose if you loose anything. Sometimes it will disable a PCIE 8x slot. And then later on if you want to use that slot you find out you can't. If you using a iATX Mini board. I would assume this might be your only option for more sata ports. And make sure you use one of the recommend models / chip-sets too. Many times some of the chip-sets are really bad preform-ace. Same goes if you get a regular PCIE card. I'll bet the NVME one though will be higher in the PCI Bus priority which will be good. But you just need to see what you might loose / gain. In my current hardware on one of my boxes. If I used that it would disable one of my PCIE 8x slots and I found it better for me to buy a HBA card and use that instead. But again you might be facing a cooling or space issue. So all depends on your case.

Well I figured out my issue. I didnt know if I wanted the extra setting to actually apply I had to turn off the export SMB under shares. Not sure why I didn't know this. I must have seen several other posts but nothing explained that.

After lots of trouble shooting other things everyone seams to think my issue is related to the SMB Extra settings. Which could be the case. I have snap shots getting generated. and I think I got my setting set right for the formatting. But Its not showing them. Here is my SMB extras Config. and an example of the snap shot file names. Also anyway to tell what user ID is used for the owner? fs-auto-snap_00-2023-03-28-1144/ zfs-auto-snap_00-2023-03-29-0247/ zfs-auto-snap_00-2023-03-29-0947/ zfs-auto-snap_00-2023-03-29-1647/ zfs-auto-snap_00-2023-03-28-1147/ zfs-auto-snap_00-2023-03-29-0347/ zfs-auto-snap_00-2023-03-29-1047/ zfs-auto-snap_00-2023-03-29-1747/ zfs-auto-snap_00-2023-03-28-2347/ zfs-auto-snap_00-2023-03-29-0447/ zfs-auto-snap_00-2023-03-29-1147/ zfs-auto-snap_01-2023-03-28-1135/ zfs-auto-snap_00-2023-03-29-0030/ zfs-auto-snap_00-2023-03-29-0547/ zfs-auto-snap_00-2023-03-29-1247/ zfs-auto-snap_01-2023-03-29-0033/ zfs-auto-snap_00-2023-03-29-0032/ zfs-auto-snap_00-2023-03-29-0647/ zfs-auto-snap_00-2023-03-29-1347/ zfs-auto-snap_01-2023-03-29-1140/ zfs-auto-snap_00-2023-03-29-0047/ zfs-auto-snap_00-2023-03-29-0747/ zfs-auto-snap_00-2023-03-29-1447/ zfs-auto-snap_00-2023-03-29-0147/ zfs-auto-snap_00-2023-03-29-0847/ zfs-auto-snap_00-2023-03-29-1547/ [HDRMovies] path = /mnt/user/HDRMovies follow symlinks = yes wide links = yes browseable = yes guest ok = yes writeable = yes read only = no create mask = 0777 directory mask = 0777 vfs objects = shadow_copy2 catia fruit streams_xattr full_audit shadow: basedir = /mnt/tank/HDRMovies/ shadow: snapdir = .zfs/snapshot shadow: sort = desc shadow: format = zfs-auto-snap_%S-%Y-%m-%d-%H%M shadow: localtime = yes

I really tied to follow all those and still no luck. if its really just the naming convention more or less then almost all the setting people have posted should be working. I dont see others changing the naming from the defaults unless they use someone else's idea. Just seams kind of silly I have to do alot of trial and error that TrueNas has figured out with perfection pretty easily. And I really dislike the guys over at XI-systems. But I'm starting to not feel very good about this either.

Any Idea how I can make the snap shots visable in windows shadow copies? I've tried several people's ideas on this and none of them are complete enough to make this work. This simple this is making me want to try migrating to TrueNas Scale. I've been waiting so long for Unraid to add ZFS and now they are close. I've never been able to make this one part work.

Ok so I have been struggling with all my recycled parts for running my server. I finally got 1 good recycled server that I think will work for my media server. Its a Dell T140 with a Perc H330 or H350 I cant remember which but either way its in HBA mode. Now I need 1 PCIE 8x slot for the HBA controller and 1 for my 10gig Nic. I would like to run a GPU in this box too but limited on lanes and card slots. Also not sure which Video card I could add. I run Dockers for downloading and Plex and Host a requester public. the CPU only has 6 cores. I got 64gigs ECC ram. Now my real questions are how should I setup my my cache drives and spin disc's. I got 12TB qty 4 drives for spin and I have 2TB SSD Qty 2 and I have 512 SSD's Gig qty 2. I dont even know if I need the 512gig ones. I was thinking maybe set those 2 up in a stripe set and use to write temp files. But they seam to over head quick if they get used much. and then stick the dockers all in the 2TB in a mirror. I don't know just trying to figure out the best way to setup with this hardware. or should I scrap this crap and do something else?

I just know here at work its all a windows shop. But the Dev group insisted on a AMD CPU server for the Data warehouse group with SQL. And they claim that AMD is much better for SQL server DB. I run a mix of crap at home. Mostly i depend heavy on my unraid servers. I tried using AMD for my backup unraid box. and I just found it to not be nearly as fast. Now I admit I'm not running Server Grade stuff at home. Its a mix of stuff left over from upgrades here and there. So I might not be the best person to reply here. But I will say out of all my reusing recycled parts and rebuilds of unraid boxes. With AMD's I generally did ok if it was doing VM's and dockers running on cache drives. But terrible at file sharing and spin disc. Didn't matter if I used HBA controller or not. Just terrible speeds. Maybe I could have reworked a few things and found a better way to use some of the drives. But with out really fine turning alot of things I just didn't have much luck on getting any where close to good speed. Now Intel on the other hand I consider it a more general over all use. Like I can run dockers and shares well. VM's do okay I feel the AMD's run them a little better. But yeah Intel does do Pass thru a bit better for VM's but I still feel the preform-ace is not as good. So each have their own. I still need help figuring out how to setup my current hardware to be more effective.

I've been getting drives here an there ready for my backup box to test this. Hope to see a RC soon.

no the dockers access the remote shares after the trueNAS VM is up. if I could get things working so Native ZFS worked and it was not as difficult. Then I might have a different way I could do all this. The Native ZFS video didnt really help with how to setup the shares and keep them working.