i-chat

im really not any expert on either the modern day docker-stuf or how swarm is suposed to operate or run. but having some aged background in it i do get how stuf can go down. just imagine: you run a simpel little unraid server that runs a couple of disks a bit of parity, a downloader a simple (probably dlna) media server and perhaps even a wordpres a minecraft or something else entirely. whatever the case may be your cpu is nearing its limits. the easiest thing to do (and probably most often done) is swapping out that motherbord the cpu and the ram, for a beefy new system runing some new 12 16 or even 24 threaded cpu, at least 64gb of ram and a total bill of sale of aproximately 500 to 1000 euro/dolar/gpb whatever. now wouldn't it be great if unraid supported running docker swarm as a controller (yes i watch youtube howto's too) in its interface. immagine going into your unraid's docker settings and selecting advanced // run as swarm controler. now when you restart the docker service it give you the proper comand to run on your alternate hardware to install docker and join your swam as workers. from that point on you aded a whole bunch of compute power. wether it be your dad's old laptop or maybe your old gaming pc or maybe that intel nuc that you got from ebay. it could now all add to your docker containers. allowing you to run nextcloud with all its features enabled or have you run all sorts of gameservers, media servers, search engines or even ai assistants where you would proviously be unable to. all without leaving the unraid ecosystem its verry userfriendlyness and the posibillity that your limited diskspace (for all those dockers and their data dirs) will soon become larger and larger (license upgrade here i come hint hint). but thats all depending on the one major question: what effect would this have to the individual containers or container sets for obvious reasons this would ever only work > if your swarm in effect still works identical to a single host/node (be it with some delay for the network trafic), > if you could still just go to apps, sellect an app, fil out its form and start it as usual.

sorry for the late reply, running the sidecar sadly didn't have the desired effect and i had a bunch mor issues with it.. i ended up using a usergroup + otp rule in cloudflare so now the tunnel only allows data if you have a username+password setup in cloudflare this provides me a simular level of protection and hase the added advantage that i can have more 'free' users for this or other apps. than i would have on a tailnet. and as a added benefit it doesnt require aditionall software or clients to be installed. if anyone is interested i could post a small howto on what i did.

what i dont get about this is: A: why change nextcloud data outside of nextcloud and even if you do, why is nextcloud AIO diferent from other nextcloud docker images? do they mount data diferently and if so do they than also allow data to be shared amoung diferent apps? would i be able still setup or use the nexcloud talk backend server somehow? one of the main reasons why im even considering this is because i kinda like to have the igpu of my pentium 5000 working for the memories app etc. @Kilrah have you any take on this?

after google'ng this error i also came to this thread so in the hopes of getting some extra understanding. my unraid version is 7.0.1 the docker container is openbudgeteer ... i want it so that this docker only listens on the ip within my tailnet so instead of having it bridged on 192.168.1.2:8880 unraid:8880 i would prefer it to only be on: 100.100.1.70 or mybudget.blabla.tailnet this would add one extra layer of security to my financials as you'd not only need to be on my wifi (and thus have access to my unraid dockers and its webapps. but you'd instead need to have access to the account alowed to login to tailnet before having to guess the webapps password. from what i understand in unraid 7 all dockers where suposed to have tail intergration but it gives this error (even when i change the docker run as priviledged (if that would 'v helped) ... please help me figure out how to fix or at least why it dont work.

after using the forum search i only saw a oneliner post on zip support, so i was ondering. i think it was / is from using one or other webhosting panel that i remeber it supporting a feature called: <upload zip and extract> i used that feature verry often to upload data folders or webapps etc to my webhosting stub. however, now using a lot of docker (including for example minecraft, apachephp8 and others where it would be realy helpfull to be able to upload entire folders or folder structures with their data) even if such behaviour would only be for a single preset / a set of aguments or a single filetype like just zip or just 7z or just .gz a user would be able to convert their archive or recreate it with their tool of choice. i would be perfectly happy for either a <upload and unzip> or a <unpack this file here> feature

honnestly i cant DISagree with this more. if you even start to think about how zfs works and what it does (and how) you would instantly find why larger pools are often "bad ideas TM" on enterprise hardware with ehterprise kind of workloads this is absolutely true, but this WILL come at an unmeasurable cost on system resources. try calculationg parity on (more than) 8 nodes double parity with a core i3 2, or 4, core cpu while also running a few services like plexserv, torrents, nextcloud and maybe a few others? when running zfs with 30 disks or more you are already looking at systems outside of the scope of most desktop cpu i fear. at least with smaller pools you can scedule them one after another untill all of them are done. i would really like to know what 'this opionion is based on, and how you percieved it to work... what in your view is the benefit ... an why / how.

this is the world up-side-down, how do you even think that, anyone should do what YOU want unless they can prove you wrong? wouldn't it be more logical if you had to prove to other people why you are right? honnestly this is just a sad way of trying to make people feel sorry for you, its just guilttripping really... allso i highly doubt that serial numbers can be securely idenified using linux cli tools... at least a tool like LSHW doesnt seem to catch every SN on every HDD or SSD. let me remind you that FREENAS world leading opensource NAS & SAN distribution only 2nd to openfiler, only recently (with version 12), changed policy about booting from USB-drives (mostly raid 0 of of 2 thumbdrives), So regaring to your 'idiotic' you really presume too much if you think that words like that could argue your point in your benifit. there you go, so let me explain, USB-drives by their protocol have a unique idenitfier that unraid uses as a way to protect its revenue for unauthories copies. it is true that some (mostly OEM like hp, lenovo, dell), manufacturers have a similar feature in their bioses that limetech could incorporate but thats only a verry small subset of hardware used by US users. So to make happen what you want, the would not only have to make rigurous changes to that base-installment system, (which admitedly can be done with medium effort), they would also have to re-invent a licensing system based of of some other unique identifier ... this might not be as easy as you think. Now this obviously COULD be done, surly, however what would be the gain. booting from sata to ram-disk is (if at all) only a verry tiny margin faster than from usb to ramdisk. and doesn't actually add much stability either, (you still would not be projected from bad writes due to disk-failure. and if you want to protect against it you would have to add some form of disk-mirroring (raid) kindly defeating the whole UNraid principle. arguably, its possible to boot of of NVME and not use ramdisk at all, making the whole system boot slightly faster and removing some of the complications that come with ramdisks... but this add a bog you your NVME disks, and might negativly impact system stability (or responsiveness when for exampe you are demaing high performance from youre write-cache ... also your reference to windows is wrong, i have been a member of the professional windows comunity for a couple of decenia now, and one of the most debated topic (complaints) agains windows it the fact that by default %alluser profiles% is stored in C:/users and not for example on h(ome):/uses (not even on windows server versions) and the only ever reponse was 'yes but legacy suport would break. all in all the cost of implementing this in my (profesional) opinion doesn't even come close to covering the win.

first of, - yes i did, only to realize it isn't readely availible here in NL, second of all its broadcom based, well lets not start on that, poor network performance with brcm chips, hickery linux drivers (even in stock firmwares), and overall it wil never suport openwrt and so will not get decent firmware updates. and even when looking for a replacement at that pricerange, ?? to compare it against: https://www.gigabyte.com/Motherboard/GC-WBAX200-rev-10#kf intel based wifi AX with aproximately 2gbps on the 5ghz band. for only half that price? the obvious question would be ... did YOU even bother.... but i will refrain from such remarks and so should you! in the end it boils down to this: is wifi worth the trouble, both in development cost and eventual support issues > you say no, (and probably have easy access to cables), > i say yes, (and do not, and furthermore, this topic has been around for years and the ye-arguments have strengthened, and the nae-arguments have weakened. for my part, if unraid was some kind of opensource software i would just have added the required software myself (or pay someone to do it) and maybe even provide build info for others. as it is - limetech can easily add the feature and NOT support it. rather than not having it at all.

after a search i think i overlooked this tread but, you are so wrong.... in another post i explain why. but basicly most (if not all) such devices run outdated mips cpu's that lack pci-e and thus the required bandwidth for gbit or higher speeds both on wifi and on lan side. read more:

im wondering if this would actually work for any 'decent' unraid aplience itself for exaple i was trying it out today on my former mini-itx desktops that i fitted with some spare harddrives. when it got time to get a trial key, nothing! - obviously it had no internet connection and i could not aply for a key. here is what i find after a long scroll through these forums. 1 you shouldn't run a server on wifi, A: why not, as i run heavy duty games as wel as conferencing and remote control software tools on it? 2: wifi is spotty, A: its true that wifi has been hit or miss over the 2 decades or so. but fact of the matter is that about the time that wifi 4 (N) was swapped for wifi 5 (AC) things changed alot for the better with WPA2 (and its linux wpa_suplicants packages) being now mature, after thair introduction in wifi N, and the new 2x2 and 3x3 mimo both speeds and stability became less of any issue. B To make an even bolder statement my wifi running a stable 800mbit at 5ghz ban 32 wifi network on 2x2 mimo with an intel chip hass about 1ms less lagg and about 10% less framedrops on hight-bandwidth protocol tests than its wired counterpart running on cat5 cables and a broadcom nic. i can only emagine what improvements i would have if i upgraded to wifi 6 (Ax) 3: you could run a cable to a 'hacked' wifi range-extender to let it act as bridge for your wifi, A: true, this could in face be done, matter of factly i have done this in the past for a 'smart-tv' that didn't have wifi --- the problems i had with this solution where that the bridge was barely powerfull enought to support youtube streaming in 4k and that the device go hot (really hot) on longer periods of action --- another large drawback is that most such devices are really exspensive or are of utterly inferior quallity. >>> most of the times you'd be lucky to find a device that even has a gbit nic, and if it has one that it is not connected internally to a usb-hub leaving it only a shared 480mbit wide bus (usb2). where'as a decent pci-e 4x4 mimo wifi ax controller could be had for about 65 to 90 and my wifi Ac card is only 45 4: go cabled its faster anyway. A Not all situations allow for a neat cabled solution, for example a rental flat where you cant (are not allowed to) drill for cable. but even if you did manage to lay down some nice cat6e cables, if you follow the news about "home IT" stuff you may have noticed that only just now the first ever true 2,5gbit switch for consumer-use was released in taiwan for aprox 90usd. by a brand called Qnap also known for it storage-solutions. ___ Mind you that its even just an UNMANAGED switch. ____ . B compare this to the gigantic leap that wireless technology has had over the last 10 years from aproximately 300mbit (wifi N ) to more than 1200mbit or even slightly over 2000 for wifi Ac and Ax respectively. In regard to the actual work neaded for a wifi main connection: 1 add wifi drivers to the kernel during build time, ar add them as loadable modules. 2 add wpa_suplicant, 3 add a simple conection wizzard - its been done over and over again and should not give any issues. 4 there is only 'one' tricky part, as wifi bridging is not well-supported. not in the 802.11 protocol no'r in most linux implemetations of any type of lvl 2 bridging. This means that any of your virtual network interface (your virtual machines) can not live in the same network as your other pc's (including the virtual-machine host itself). so connection your VM to the same router as the host will not easely be possible. A: however if you google on the matter there are some ways to trick all parties (computers) in thinking that its atually still done. (google: wifi bridging kvm) B there is also the option of inter-network-routing: meaning that you have to tell your router (and its routing table) that there is a another network with all new ips behind your vm-host's ip-adres. your vm host would than have to act as an osi-model level 3 router. C the least eligant but easiest setup hover is to not try at all and instead use a tunned interface between your vm's and the router as allmost ALL home routers support some sort of vpn-module with settings to let the remote users share the 'home ip range' all your issues would also be solved at the cost of a slight to moderate performance hit. IF you opt for option C however, i would strongly advice you to losen your security setting to near zerro for the encryption-settings. as your vm-host lives on your home-network there is hardly any use (unless you really want to) to add more performance hit to your connections. this is true espeially on the router-side because en- and de-crypting the data from your virtual network connetion is the hardest part for its tiny cpu. Unlike modern intel- and AMD- CPU's most mps-based routerbords do not support hardwarebased AES-decrypting. So far my 2020 'updated' view on wifi can, and should be supported.