awediohead

Thanks very much! I feel I can go ahead now with a lot more confidence that I'm not about to do something dumb (again!) I think I'll keep it simple and just swap the NICs to start off with. I should be able to add the LAG afterwards once the unraid box is back where it lives next to the switch and router. And yes it is complicated - I have reams of notes and spreadsheets about how the switch, router and unraid are theoretically going to work together. Makes my head hurt, but kind of fun in a wirdly masochistic way cheers

I've currently got my server in my office trying to figure out some stuff in the BIOS (about C states and power efficiency) which means I have a monitor and keyboard directly attached. So it's an ideal time to change from using the onboard Realtek NIC to a four port Intel PCIe card I installed a few weeks ago. I had a go at this last night and made a total cock of it. I ended up being unable to connect to Unraid Connect and unable to ping even Google. Having done a bit more research I think I understand that it's generally a bad idea to have two NIC's active at the same time, as in two pieces of hardware. Whereas a multi port PCIe card should be OK . . . is that about right? I think it is, because I only managed to get Connect working again (by signing out and back in several times plus restarts) by undoing all the changes I'd made to the 4 port NIC's settings and completely reverting to the Realtek NIC. Meaning currently all four ports on the Intel NIC are marked as 'shutdown (inactive)' in blue. So what I think/hope will work is: 1) Go into the BIOS and turn off the onboard NIC, boot into Unraid. Stop Docker Services and VMs. 2) Go to Network Settings and check that - in the absence of the Realtek NIC - the Intel ports are now eth0 to eth3, when they're currently eth1 to eth4 3) Enable bonding between eth0 and eth1 and also bridging - I plan in future to have an LACP based LAG between my switch and Unraid - though right now I can only connect to eth0 via a dumb switch because my managed switch is in the server closet. 4) Enter the static IP I have assigned to the Intel eth0 MAC address in my router, so this IP becomes the new admin access IP for the Web UI. 5) Restart - maybe sign out of Connect and back in again. Is it maybe a good idea to sign out of Connect as a step 0) in the above list? Any pitfalls and problems I'm not seeing? Any other suggestions? Advice? Thanks BTW longer term the plan is to use eth2 passed through to a VM and eth3 as a dedicated port for security cameras - though currently I don't need to run a VM and haven't bought a single security camera, but that's the vague plan ATM. PS If I change the IP assigned to the server will Unraid 'automagically' manage changes to the docker app to host port mappings? i.e. change for example the current 172.17.0.2:22000/TCP <----> 192.168.1.xx:22000 for Syncthing to 172.17.0.2:22000/TCP <----> 192.168.1.yy:22000. ? Cheers

I can assure you I'm not an AI bot. As I posted this a while ago I've been digging a lot deeper into how to do what I want to do and the first steps involve ignoring Unraid and focusing on getting pfsense and my switch set up correctly. Then I'll circle back and think about Unraid. Thanks

I don't understand what you mean by "just lost the internet to the server'? Do you mean your other home network devices can connect to the internet but not your unraid server? If that's the case look to your router for the problem not unraid. Unless of course your running router software such as pfsense virtualised on unraid? BUt if you've done that I don't think you'd be asking these questions. Are you trying to access your server remotely or are you on the same network? If you're on your local network internet connectivity is irrelevant to your access to unraid. If your ISP is changing your WAN IP address - which happens to everyone unless you pay for a static WAN IP - this should have zero impact on your local network IP's whether they are assigned/static/reserved or not. The functionality of a PC's power button is configured in the UEFI BIOS and if you've updated the BIOS then this could have been reset to a default value that won't allow a hard reset. The power switch could also just be broken - either way your power button's function has nothing to do with how your network functions and whether or not you have an internet connection. When you assigned local network static IPs in your router did you change the DHCP range? For eg if you have a 192.168.1.0/24 LAN network then normally your router would be 192.168.1.1 and DHCP will fairly randomly assign IP addresses to LAN devices from 192.168.1.2 to 192.168.1.254. To use static or reserved LAN IPs this DHCP range should be changed to exclude ranges from which you select the static IPs. For e.g. You might set the DHCP range from 192.168.1.100 to 192.168.1.200 so that you can assign static LAN IPs from 192.168.1.2 to 192.168.1.99 and from 192.168.1.201 to 192.168.1.254 so they will never conflict with DHCP assigned IPs. Different routers handle this more or less intelligently and - of course - if your router is getting 'confused' when your ISP changes your WAN IP (which should never happen but does) then resetting the router or checking for a router firmware update might be a way forward? I don't know but I suspect that what isn't making sense is perhaps because you're trying to solve a router problem by making changes to unraid?

Can anyone tell me if the new 6.12.4 specifically addresses hard drives spinning up for no apparent reason? Wondering whether to upgrade now or (as I normally do) wait and see for a few days. It seems to happen even if I have turned off all dockers as well as VMs and when there are no active shares. In normal use I would expect the cache drives to remain spun up as well as whatever drive was being accessed by Plex streaming - however for the most part it seems they all spin up - even an unmounted unassigned device. I have read through numerous threads where this has been reported by others going back several unraid versions and tried some of those 'fixes' - e.g. turning off all dockers and gradually turning them back on to try to identify the culprit - but as it happens when they're all off . . . also switching off temperature sensing/reporting by unloading the drivers of the System Temp plugin. When I manually spin them down it seems to quite cleverly spin down the not currently in use drives, but then next time I check as little as 10 minutes later, they're all spun up again. Hope someone can help. Diagnostics attached singthesis-nas-diagnostics-20230911-1338.zip

Thank you very much for both your suggestions and sharing your experience with LAGs I have no idea whether I need a LAG or not. Part of it is just that I have these ports doing nothing so thought to put them to some use. The other part is that, with my partner streaming from the server, and possibly a couple of other family members doing so, simultaneously with me footling about with unraid, I thought a LAG would just provide the extra bandwidth to help everything go smoothly. It's very likely that with everything now connected by ethernet instead of the crappy wifi we've been dealing with since we moved house last year a LAG would be overkill. Also since I first posted the question I've realised that it probably makes more sense to do something even further beyond my current understanding, which is to use the different ports on the Intel PCIe card for different roles instead of aggregating them, by creating VLANs: One for regular internet access, Plex streaming and SMB shares to family PCs, one for my admin access (so my admin access to Unraid, the switch and my router are all on the same VLAN to be more secure) and one to pass through to a VM. This of course begs the question as to how one PC (mine) can access multiple VLANs and I think that's possible without a Layer 3 switch if my PC goes direct to a router running pfsense instead of the current off the shelf router or the layer 2 switch. Much more to learn and research!

So I recently finally managed to complete installing ethernet in my home. We now have two ethernet cables going to three bedrooms and our lounge, (8 runs in all) with the unraid server, router, switch and patch panel in a centrally located closet space. Because of my own and my wife's disabilities our bedrooms are also where we have our personal PC's. As a housebound person she makes fairly constant use of this PC, to entertain herself, distract herself from chronic pain, shop online and generally communicate with the outside world. IOW my messing stuff up goes way beyond being just a bit inconvenient. Right now everything is working OK, so any changes I make going forward need to be carefully planned to minimise downtime, especially to internet access, but also to the unraid server which she uses mostly for Plex/Jellyfin. For e.g. I have a power efficient little S920 Fujitsu PC that will make an ideal router to run pfsense on, but for now I've just continued to use an Asus RT-AC86U router which is also the wifi AP. The plan is to switch this to AP-only mode when I eventually replace it with the S920 until I can afford a dedicated AP, but I really need to get through the basic tinkering, learning and making mistakes stage with pfsense BEFORE swapping it out. In the meantime I'm doing a lot of watching videos and reading up on how other people do things but since everyone's hardware is different, while I am learning a lot, I'm also feeling pretty overwhelmed with all the variables to consider. If it were just me I'd play around, experiment, test, break things and reinstall. Not an option if it'll bork my wife's internet access or access to the server, though the former is very much more critical to her. So, some the variables are: Hardware: S920 has 5 x 1Gb ports, Unraid server has 5 x 1Gb ports - both have a four port Intel PCIe card + onboard Realtek port. The Netgear switch has 16 ports - using 8 from the patch panel (1 to 8 - two runs from each of the four rooms) with port 16 going to the RT-AC86U. Software: VLANs and LAGs, Firewall rules Tutorials online talk about the importance of having a management network to administer the server, router and switch that'd also need to be accessible by my PCs, but not from other family PCs or devices for security. So my first question is how would I go about having a separate admin network/interface to access the unraid web ui, the router web ui and the switch web ui, without interfering with my own or my family's access to SMB shares, Plex, and the general home network functionality of other docker plugins. For VM's I'm assuming I'd pass through a dedicated port - but then how does it work that from the same PC in my room I'm able to access the various Web UI's, and watch Plex (or similar) and access a VM if they're all on different networks or VLANs? At the moment I can do all that because it's all on the same network of course. Ideally I'd like to self-host a small website and allow access to some aspects of my server to close family and friends, so I don't want to take shortcuts with basic security practices for the sake of convenience. On the other hand applying the Keep It Simple Stupid principle makes it seem that I'm unnecessarily complicating things. Sorry for the wall of text - I've a lot more questions but some pointers on how to get the foundation / fundamentals right with planning would be hugely appreciated. Thanks

I've no experience of NordVPN specifically but with Mullvad there is a setting that allows network communication outside the tunnel to local multicast and broadcast range as well as to and from these private IP ranges: 10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16, 169.254.0.0/16, 0xfe80::/10 and 0xfc00::/7 I was only able to access my local network generally and SMB shares once I had selected this option in the Mullvad settings. Possibly there's a way for you to 'white list' the above?

I'm probably not the best person to help, but until someone else replies . . . the more info you give the better people who do know can help: So far we know that you have a MacBook Pro and that the PC you're trying to install Unraid onto has an Intel CPU and has some 'clean' hdd's. That's not a lot to go on . . . In general with Linux I tend to turn off the "windows" related stuff in UEFI BIOS - I've no idea if its necessary these days, but it used to be. This is mostly around turning off Secure Boot, selecting 'other os', turning off CSM. I also personally turn off XMP in RAM settings as you typically want reliability over speed in a server. Then you want to make sure the USB flash drive Unraid is on is the default boot drive and this typically means changing the default order which is going to be assuming you want to boot from an SSD or HD not a USB drive. That's all I can think of for now. Good luck

Just struck me that with so many having viewed this thread and no replies you're probably all thinking, "just figure it out with a bit of trial and error" - which is fair and in normal circumstances it's exactly what I'd do. However, my partner is severely disabled and very reliant on our media server for distraction from chronic pain. She can manage without for a half hour or so with a bit of forewarning, but she can't manage with me borking the whole thing for hours or days because I've done something dumb, especially if there's something I can do to avoid it - which is why I asked for help in advance here. Hope someone can offer some guidance. Thanks

So I recently picked up an Intel 4 x 1 Gbit NIC off eBay and I was wanting to start using it instead of the Realtek NIC on my motherboard. I'd really appreciate some guidance on the steps I should take, and the order in which to do things (which I suspect is critical) so I end up with a two port LAG from Unraid to my Netgear managed switch. Is it preferable to just go with setting a new known static IP address for the new NIC or try to re-assign the existing static IP address - which is obviously associated with password managers for multiple docker logins etc. and possibly multiple system settings in Unraid which might (?) get broken if I change it? Obviously I need to somehow try to retain a connection to the Unraid Web UI through the process of making the change. This is as far as I've got: 1) Set up a LAG on two currently unused ports on the switch - just RTFM'd the Netgear GS116Ev2 re LAG and learned: "The switch does not support IEEE 802.3ad Link Aggregation or Link Aggregation Control Protocol (LACP) groups but supports manual static LAGs only" - I've seen mention of IEEE 802.3ad and LACP in other forum threads, but I don't understand what NOT having them means to me? 2) In Unraid network settings where eth0 is the Realtek motherboard NIC, I create a bond between eth1 and eth2 on the Intel 4 port NIC and also enable bridging. 3) Connect the eth1 and eth2 ports on the Intel NIC to the switch ports defined in the LAG (which happen to be ports 13 and 14 for my reference) with a couple of cables . . . I assume at this point my router will assign via DHCP an IP address to this LAG as if it were one interface. (?) Also that because it obviously won't be the IP I access the Web UI through which is tied to the MAC address of eth0 that at this point I'm still able to access Unraid WebUI 4) No idea what to do next ? . . . I just deleted three paragraphs of speculation about what might go wrong before realising it was pointless! Any help greatly appreciated!

Title pretty much says it all. I have 1 x 4TB HD parity, 4 x 4TB HDs, 2 x 500GB SSDs, one of which is currently unused, one is the cache drive for most shares, and 1 x 8TB unassigned device drive, and the flash drive of course. Sometimes things seem to be operating as I'd expect with the cache and maybe one of the array drives having a green light on the Web UI, because someone in the house is streaming from Plex, the others being grey to indicate they're spun down. However, for no reason I can fathom, five minutes later, when I check the Web UI main page all the drives are lit up green when nothing else (I'm aware of) has changed. Why the unassigned device should spin up, even when some other automatic process is spinning up the array, is weird. In the disk settings I have them set to spin down after 15 minutes, but they don't spin down. I've tried resetting this (the changing the setting to 30 min, applying and then changing it back again trick I read about on these forums) and the server was rebooted yesterday. Is there a setting I'm missing? Maybe even something I need to change in the UEFI BIOS? Generally very keen to learn about anything I can do to reduce power consumption and running costs. Thanks

I can't say I've "solved" this by a long stroke, but I thought I'd post back here to say I found this thread useful https://forums.linuxmint.com/viewtopic.php?t=318242 specifically this: Though this mentions Nemo, I'd actually wiped Mint in favour of Pop OS by the time I found this, but the same applies to Files which is the Pop OS version of Nautilus So what worked for me was to go to +Other Locations and totally ignore the supposedly handy direct Network links to my unraid server and other devices on the network. At the bottom of the Files window there's a field with "Enter server address" into which I put smb://nameofmyunraidserver.local/nameofshare I find it very odd that these things don't just work OTB between two machines running Linux (NFS didn't work either) and all the tutorials I could find target setting up Samba to share a folder on the machine you're on, rather than to connect to a share on unraid. If I've just missed a straightforward quality tutorial despite hours of searching please post a link! This one from Matt Hartley was the best I've found so far: https://glovesoff.substack.com/p/samba-on-linux-the-easy-way

So I'm getting this red warning on 6.12.0 stable do I just delete the line in /boot/config/vfio-pci.cfg ? The line is: BIND=0000:06:00.0|10de:1d01 0000:06:00.1|10de:0fb8 I think it might be because I removed a GPU a while back but as I'm not currently running any VMs and this only shows on reboot . . . The file is otherwise blank apart from this single line - just need a little reassurance before I delete something I shouldn't!

I've just installed Linux Mint on a laptop we helped a completely non-techy friend buy. She's a close friend who cares for my disabled wife and the idea was that she would occasionally use our unraid server to back up various files when she visits. I have done a lot of searching online to try to resolve this before posting here but I've found nothing that makes any sense as to why this shouldn't 'just work' especially since it's a fresh install of Linux Mint. So on MacOS Finder I can easily (as I have for years) just enter the user share name and password when I get the 'connect as' prompt in the Finder. It immediately connects to my unraid NAS shares and lists the available folders. When I try to enter the same credentials in Linux Mint's file browser under Network -> NameofmyNAS (File Sharing) I get "Unable to mount location - Failed to retrieve share list from server: invalid argument" In the past I've quite easily managed to access my unraid shares via my own Linux Desktop install which is a Fedora spin - and the searching I've done provides "solutions" that are so very far beyond my understanding and refer to a bug in gvfs not Samba on Ubuntu or talk about editing various conf files in the terminal - none of which should be necessary for something I'd have thought was pretty straightforward? I suspect that once again something really fundamental in my understanding of how Linux works is the failing here - but for something to work flawlwssly on MacOS and not another Linux distro that's supposed to be super beginner friendly seems weird. cheers

Thanks Conner. Yes I should have said that I'd already set the BIOS to "Typical Idle Current" when I updated the UEFI BIOS - I agree it does seem to be THE critical factor since I'd never got around to changing the RAM speed from default when I was having multiple problems. As I said before, when I ran Memtest86, it defaulted to running it at 3200 MHz and my thinking was that if it passed with no errors at a higher clock then it should be fine at 2133 MHz for unraid. TBH the leap from a 4C/4T Haswell era CPU to Ryzen 3rd gen with double the RAM I had before is so enormous I'm not bothered about 2133 vs 3200 so long as everything works at least as well as it did before. Thanks for the clarification on JEDEC vs XMP vs anything in between.

Maybe someone can clarify what the linked post means by "overclocking" RAM exactly ? - is it saying that any XMP/DOCP profile should be totally avoided or is it saying that RAM speeds should be limited according to the tabulated numbers for each generation / memory rank? I'm confused because my RAM's basic clock is 2133 MHz but was bought as 3200 MHz RAM (2 x 16) and the 7 hour Memtest86 test I ran also defaulted to testing it at 3200 MHz with no errors. For example on my desktop PC running a 3700x on B450 I have 3773MHz RAM (2 x8) so if I were to use that with unRAID I'd understand it would make sense to underclock it to 3200 MHz in line with the settings listed in the tables. I'll be sticking with 2133 MHz for a while because I can finally get a reliable connection over LAN - the new Connect version of Myservers seems to have solved some problems I was having!

Thanks Hoopster, you posted the same link about RAM on Ryzen that I'd already linked to - first line of second paragraph of my post For sure it mentions some other things but I'd meant to make it clear I was looking for any info further to what was mentioned in that post, especially since that info was current 3 years and many BIOS / AGESA revisions ago. Happily memtest completed with 0 errors at 3200, so from that can one assume that there will be no errors at the basic 2133 MHz? Or is that a false assumption? I've been having all kinds of hassle with the My Servers plugin reporting my server to be online when I know for a fact it's switched off. So it's perhaps not that surprising that when I try to use the "Local Access" link it fails to connect even when I know it is switched on. That's meant I'm juggling graphics cards trying to get access to the GUI directly to make changes before trying it via my LAN again - rinse and repeat. I ended up going down a fairly lengthy rabbit hole on account of Firefox and Brave both auto-correcting my unraid IP from http:// to https:// everytime I try to login - that might have been resolved by deleting the cache and site data (everything) from the privacy / history preferences in Firefox, but it's 1 am and I'm done for today. cheers

So I've just switched from my old haswell based z97 motherboard to a Ryzen x570 + 3600 + 32GB 3200 MHz RAM with a GTX 1050 for transcoder duties . . . . and things haven't gone well so far. I definitely hadn't factored in the downsides of not having an iGPU when you can't login over a network - I think this is now solved and was because I'd set up my router (asuswrt-merlin) to bind the assigned IP to the MAC address of the old motherboard NIC. I've read the info here about not overclocking RAM on Ryzen, and I'm currently running memtest86 via a Ventoy flash drive. We'll see what happens! The hardware was previously multi booting as a hackintosh with windows and Fedora without issue, but the RAM is a new variable. Any other Ryzen x570 + 3600 tips and tweaks I should be aware of? - such as specific BIOS settings? CSM enabled or disabled for e.g.? Switch off all the secure boot stuff and select "Other OS" ? Can I assume that selecting all the virtualisation options (SVG, IOMMU etc) are unlikely to cause problems for unraid, while more "gaming" settings like re-sizable bar should be disabled because they're not used? Or are they? I hadn't previously updated the BIOS/UEFI when I was having problems last night, so I'm hoping that solves a few issues - I can now at least set it to consider usb flash drives as drives, which wasn't an option with the earlier BIOS. I have two GPU's available: the Nvidia 1050 mentioned earlier for transcoding plex etc., and I also have an itx version of an rx570 4GB which I'd planned to passthrough to a MacOS VM (now I have a few more threads to play with) - however in this testing and troubleshooting phase would I be better off just using the rx570 and removing the 1050? That question assumes Linux generally plays nicer with AMD over Nvidia . . . . Or could that cause problems because the unraid config is expecting to find a 1050 that wouldn't be there? Any other pointers gratefully accepted. Cheers

Hi - as you can see there's been no informative responses to this thread so I kind of gathered the idea was an non-starter. That said I also shelved it because I realised that with my current pretty old hardware asking it to be a media server and run a VM meant it would do neither very well, while it is a perfectly adequate media server. I'm planning on upgrading my hardware to a more modern CPU and motherboard and may revisit the idea then, but to be honest I've had such good results running MacOS bare metal on hackintosh builds, which also easily multi boots other OS's, it would probably take a dedicated "Make your own DAW on MacOS VM" video walk through from Ed of the SpaceInvader YT channel to reassure me that my efforts wouldn't be a collossal waste of time. I have heard good things about using Proxmox to this end for video editing with FCPX, which would seem more demanding that just audio pass through. If you have a desktop type PC and can add a dedicated PCIe USB card to pass through for both audio and the MIDI controller I'm pretty sure it's doable with whatever hypervisor you use. HTH

I tried to upgrade to 6.11.0 a while back and had a lot of problems so reverted back to 6.10.3 - which had been stable for months previous to the upgrade attempt and has been since I reverted. At the time I assumed this was probably because my hardware is so old (z-97k i54690k 16GB RAM DDR3 1600MHz) - so I'm just wondering if any "older hardware" bugs have been identified and sorted since 6.11.0 that should give me more confidence to try to upgrade to 6.11.5?

thank you both very much - again! MAM59 - I have a very current Memtest USB flash drive I will definitely do some tests with. Excellent advice! I have a couple of kits of DDR4 RAM to test - one 2 x 16 at 3200MHz and one 4 x 8 at 3733 MHz so it'll be interesting to figure out what works best. Frank1940 - I have some VMs installed but 99% of it's current usage is as a media server with dockers - with just the 4 cores and no multithreading it's just not pokey enough to do media server duty and run a VM well at the same time. Given that I only ever installed the VMs to play around with and once they seemed to be working OK, haven't used them since, I'm happy to just delete them. I can always reinstall them if needed. Cheers

Thanks MAM59 and Frank1940 Murphy's law, computer tech and I are long time acquaintances - so I appreciate the heads up on settings. TBH I've been putting this off for months because things do tend to go pear shaped on me, but forewarned is forearmed. With regard to not overclocking RAM - I assume 'overclocking' includes any kind of "XMP" (I forget what AMD calls ii) type "overclock" - so to just run it at the default JDEC speed? I'll have a dig in the BIOS UEFI settings for C-states too of course, and I assume the EFI MAM59 mentioned is basically about figuring out whether or not its currently UEFI or Legacy booting because it will be UEFI booting on the newer hardware? Bottom line is there's a lot of people here who would either have a very solid plan in place before upgrading hardware, or are already so knowledgeable they wouldn't be very phased if things went wrong. I can only try to do the former, so I really appreciate the guidance. That's especially true because my wife's disabled and very reliant on the server for entertainment and distraction from chronic pain. It's a coin toss between leaving it well alone, on an "if it ain't broke . . . " basis, and accepting that 9 year old hardware won't last for ever. Cheers

Really just looking for any tips and advice to help the upgrade process go as smoothly as possible: For example, since I'll be changing from an old Asus z-97k i5 4690k 16GB RAM DDR3 Intel based set up - to an AMD x570 + Ryzen 3700 with 32GB DDR4 - is there anything I should be watching out for, or anything I should "instruct" the system to do as it restarts to make it recognise the new hardware? Or should this all just happen " automagically" ? Things "just working" has generally been my experience with using various Linux DE distros between different PCs, at least after a couple of restarts and updates, but I wanted to double check. I'm not planning on making any initial changes to the array and cache drive ( 5 x 4TB drives on the array) and 2 x 500GB SSD cache drives, but should I be trying to copy over what order of SATA ports/connections each drive is connected to? Or are the drives recognised and ordered by their UUIDs? In other words when I start up the new build will UNRAID identify the drives in the array from what's stored on the UNRAID USB boot flash drive? Anything else I should think about? Prepare for?

If you're saying that you had four drives on your array, presumably one of them your parity, and the fifth drive was an unassigned device then you should be OK. You might choose not to even connect the unassigned device until you're sure your array is functioning properly. I'm pretty sure (though no sort of expert!) that you would not want to attempt to make up your "array" by mixing SATA connected and USB connected drives. I'm just in the process of figuring out what I should do to prepare my unraid set up to upgrade the hardware from a fairly ancient Haswell era board and CPU with DDR3 to a much more modern x570+3700 set up - which is similar enough to your question for me to be watching out for better informed answers than mine!