The Class E Network

    • Offizieller Beitrag

    RE Webmin and the default Samba setting:


    The samba version is (4.8.4) . It's current. The running assumption almost has to be "default" = "negotiate", which starts with the latest (SMB3.0). Otherwise, updated Windows clients wouldn't connect to it.
    ___________________________________________________________________


    XFS doesn't handle power outages all that well either.

    XFS, ZFS and others are examples of LAN administrator/Server Farm file systems that are almost guaranteed to be run on an UPS. This type of user and environs are not what OMV is targeted to.


    It will keep changing. Trust me. That is why I haven't ported all of the plugins.

    It will use buster. And Volker won't release it before buster 10.1, I'm guessing. So lots of time for changes.

    So, in so many words, are you suggesting that the "BTRFS only" approach is still on the table? From what I gathered from the discussion thread on GitHub, there are more that a couple of dissenting opinions, about this, from experienced users.


    I'm being honest when I say, I "like" the ideas behind BTRFS. Really. There's kernel integration (automated upgrades), low resource requirements, and there's the novel ability to shrink a RAID array among other unique and novel features. For these reasons, when looking for something for bit-rot protection, BTRFS was my first choice. I even signed up for the projects mailing list and sifted through a chunk of their errata. My sum total impression was, "Not Ready", but I gave it the benefit of the doubt and tried it in a single disk setup. The results? Not so good. Two serious failures, with a device that's not on-line very much.


    These things appear to be obvious to those who really look at it:
    BTRFS is not fully developed, the CLI utilities are still inadequate and even the experts don't really know how to "Rx" a busted file system. (I became painfully aware of this when trying to fix BTRFS filesystem issues.) Again, more votes for, "Not Ready".
    _______________________________________________


    - Let's just assume that BTRFS doesn't like power outages. (My anecdotal experiences are certainly not empirical.) Now, let's look at OMV's target user base which consists primarily of, "small businesses" and "home users". As previously mentioned, I'm going to estimate that thousands of these users are running without an UPS.
    - Given OMV's user base - NOOB's and non-LAN admin types; the base file system requirement should be heavily biased toward stability and reliability. Along those lines, EXT4 is proven - BTRFS is not.
    - Just because BTRFS is easy to write code for, doesn't negate the basic requirement for stability and reliability.


    I can easily see where vfrex's comment about setting up better support for BTRF makes sense. Add support for BTRFS in the GUI and see what happens, as a sort of BETA test. But jumping into BTRFS, without broad based experience and testing, to the exclusion of all else, seems like a reckless move. There's no reason to rush this and risk the loss of a substantial percentage of the user base.


    I'll get off my NOOB'ish soap box now... :)
    I realize the final decision is Volker's and those who are working on this project (unpaid) have the liberty to do as they please.


    On the other hand,, well,,, I guess we'll see what happens...

    • Offizieller Beitrag

    XFS, ZFS and others are examples of LAN administrator/Server Farm file systems that are almost guaranteed to be run on an UPS. This type of user and environs are not what OMV is targeted to.

    xfs is actually the default filesystem for Redhat/CentOS even on desktop systems and laptops
    .

    So, in so many words, are you suggesting that the "BTRFS only" approach is still on the table?

    Not sure. I don't get to make that decision.

    Just because BTRFS is easy to write code for, doesn't negate the basic requirement for stability and reliability.

    I don't think he would pick it because it is easy to write code for. It is the only option to write code for since he doesn't like the zfs license situation.


    BTRFS is not fully developed, the CLI utilities are still inadequate and even the experts don't really know how to "Rx" a busted file system. (I became painfully aware of this when trying to fix BTRFS filesystem issues.) Again, more votes for, "Not Ready".

    I'm not disagreeing with that but there are commerical NASes using btrfs.


    Hopefully, Volker will read it and at least give it due consideration

    That is all we can do :)

    omv 7.0.4-2 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.10 | compose 7.1.2 | k8s 7.0-6 | cputemp 7.0 | mergerfs 7.0.3


    omv-extras.org plugins source code and issue tracker - github


    Please try ctrl-shift-R and read this before posting a question.

    Please put your OMV system details in your signature.
    Please don't PM for support... Too many PMs!

    • Offizieller Beitrag

    xfs is actually the default filesystem for Redhat/CentOS even on desktop systems and laptops

    Redhat :) ... XFS,, coming to us from the same folks that depreciated BTRFS. (Can you see a bit of irony in that? :) )
    Still, XFS has a longer history and performance record when compared to BTRFS.


    Laptops and Desktops, from a file system perspective, are two distinctly different devices. One has an UPS built in, the other does not.
    ____________________________________________________________


    Along these lines, it seems we've come full circle on file systems. Back in the days of Windows 3.X, (FAT) it was possible to lose power and end up with a system that wouldn't boot again. (It happened to me once.)
    Then effort went into writing journaling file systems that could survive dirty shutdowns, with processes built in to "clean" the filesystem (if needed) on the next boot. A good degree of fault tolerance has been achieved, over the years, and EXT4 is a good example.


    Now the marching banner seems to be integrating volume management into the file system, drive aggregation (RAID like options), snapshots, etc. Interestingly, as the power grid ages and gets worse as time goes on, fault tolerance is falling to the wayside as the need for it is ever increasing.
    With the baton handed to a new generation, it appears that we're doomed to learn the same lessons, over and over again. Power and fault tolerance, "dull and boring" where new features have "spark and glitter" and are exciting. Why should fault tolerance hamper the development of the new and improved? (We'll never learn.... :) )

    • Offizieller Beitrag

    After a few weeks of waiting for the adapter, and a bit more time to dig the box out and install a quiet heatsink/fan combo, I finally got my home theater i5 on-line.
    (I still would like to know "why" Dell believes on using a 5 pin fan connector is necessary. Give that only 4 pins are connected, it seems overkill.)


    It's somewhat amazing how much CPU power is required to smoothly stream and render flash and HTML5 video. I'm sure that a variety of factors are involved, but it seems a Win7 Core2 duo couldn't do a smooth job of rendering. If any additional task kicked off, virus scan, etc., that was "it".
    ______________________________________________________


    @geaves :
    I added a new sub-section to the Win10 Connect How-To. It's possible to bust Win10 Clients out of a domain, to connect with Samba shares on external (non-domain) servers and it can be done with a single registry or secpol setting.

    • Offizieller Beitrag

    It's possible to bust Win10 Clients out of a domain, to connect with Samba shares on external (non-domain) servers and it can be done with a single registry or secpol setting.

    Yes, but that would have to done by a domain admin and to to do something like that it would probably be set using group policy. According to the document NTLMv2 is already predefined in server 2008 R2 and upwards, so I would expect that to be set on a client machine to ensure authentication to the domain.
    Users such as my wife and daughter have laptops they use 'away' from the domain either at home or at another location in relation to their work, that is because a 'roaming profile' has been set on the domain, so to login to the laptop they use their domain user name and password. This ensures that GP's are still in place if the laptop is away from the domain.
    The only way a user could change that setting is to be able to log in to the laptop locally -> admin\<machinename> -> password, so unless they know the local admin or admin user name and password they cannot log in to the machine locally to make those changes.
    So if it's predefined on the server it would make sense for that to be enabled within GP for each domain user. But that leads me to 'why is it necessary to locate your server by IP and not hostname'. Well in the above scenario it's relevant because their laptop's are part of a domain, they are not part of a home network, but they do obtain an IP address from my dhcp so they have access to my network. So instead of using \\<hostname> to locate OMV they use \\<IP address of OMV>. Something I did learn the other day, I have a network printer and W10 automatically installed the printer on their laptops once they were connected. Whilst W10 can be a PITA W7 would never had done that you would have search for the printer and install it, which W7 would not allow unless you had admin rights either as a domain admin or local admin.

    • Offizieller Beitrag

    The only way a user could change that setting is to be able to log in to the laptop locally -> admin\<machinename> -> password, so unless they know the local admin or admin user name and password they cannot log in to the machine locally to make those changes.

    Well, I'm sure (I know) that it's unlikely that a domain admin would make that change or let a user make a policy change like that - even if the platform is a quasi-personal device an employee can take home. But there are interesting admin possibilities on the server side, where merging a Linux Server into (domain) Computers and Users can be problematic at best. (More than one admin has asked this specific question on the forum.) With the barriers in place, to prevent connecting to unknown or rouge servers, I find it interesting that a single policy change can bypass it all.


    BTW: you're right about the core of this being a secpol.msc change. In fact, the appropriate change in security policy creates the registry key used in the How-To. I thought it best to go direct with registry key creation, at the highest security level. It's terse, task oriented, and done. If users go poking around in in sec pol, in the GUI, it could result is serious problems.

    • Offizieller Beitrag

    We're setting here on generator power because of high winds. (And, presumably, downed trees somewhere.) Power went out yesterday afternoon, when the winds where absolutely howling. The phone hung in there, with the DSL, until late evening then it died as well. This morning, there's still no power but at least the phone and DSL are back, hence this post.


    Our 60 Grit friend, in another post yesterday, said that one of his friends (who works for a "real big" agency :) ) did away with UPS systems. 60 Grit thinks this is "progress" and now, apparently, running (BTRFS) without an UPS is OK. I suggested, if doing away with UPS (maybe generators?) was his friends idea, he should keep his resume up-to-date.


    I suppose if one lives within a fairly small area, is 20's to 30'ish and hasn't seen line hits or power outages before, it might seem that way. (That an UPS is an unnessary expense.) Talking about complex systems and certain inevitabilities would have been pointless - he has no experience or context in the area. But the statement; "the grid works, until it doesn't", is easy enough to understand. Decent UPS systems and power generation have never been cheap, but doing away with them is a lot like throwing out all equipment fuses because you've never used them before. When needed, they're worth their weight in gold.


    I've seen line hits destroy equipment and when power go out in climate extremes, tropical or arctic environ's, it's no longer a professional consideration. It devolves, quickly, into an emotional issue. Heat, Air-con (and IT :) ) are not an option.
    _________________________________________________


    What can one say? I'm increasing of the belief that humanity is doomed to learn the same lessons over and over, generation to generation.

    • Offizieller Beitrag

    It doesn't matter how good your grid is if someone accidentally switches off the wrong breaker :)

    omv 7.0.4-2 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.10 | compose 7.1.2 | k8s 7.0-6 | cputemp 7.0 | mergerfs 7.0.3


    omv-extras.org plugins source code and issue tracker - github


    Please try ctrl-shift-R and read this before posting a question.

    Please put your OMV system details in your signature.
    Please don't PM for support... Too many PMs!

    • Offizieller Beitrag

    It doesn't matter how good your grid is if someone accidentally switches off the wrong breaker

    I agree. "Stuff happens". This the very concept behind insurance, and an UPS is "tech" insurance. Speaking of "breakers", back in the day, something like that was called a "CLE" (a Career Limiting Event). :)
    ___________________________________________________________


    Of power and CLE's, I saw a CLE in action - in bold living color.
    Me and my bud were getting ready to leave the sandbox and this guy, who I'll refer to as "skippy", was coming in.


    The compound had a 100KW generator installed, without a transfer switch, and most of it's safeties by-passed. (This was not my doing, it was cobbled together by a Military type.) "Skippy" was briefed on the situation and told, in no uncertain terms, that the generator was NEVER to be fired up without the line MAIN shut OFF. A day or so passes and on the very day when me and my cohort were scheduled to leave, skippy fires up the generator but leaves the MAIN ON. And,, city power came back... Apparently there was a war, between the line and the generator, that no one was aware of until it was way too late. As no surprise, the line won. Man, that generator reeked! I'm sure you've smelled burnt windings, but this was on a scale I've never seen (or smelled) before.


    One of the smaller aggravations for my bud, is that he had cloths washing at the time, in prep for packing and leaving that evening. They were soaked. No power, no dryer. We rocked his cloths down to a picnic table, knowing they'd dry out in the arid air in a couple hours. (Otherwise it was wet cloths in a trash bag.) But,, there was a mild dust/sand storm progress as well. :) (Common for that location.) While it dried, his stuff was so gritty, he had to bag it anyway. Man, he was fit for a straight jacket. :) I ran some intervention to prevent him from kicking "Skippy's" axx.


    Definitely, a CLE. (And a shame because that sort of stupidity tears down healthy working relationships with other entities.)


    we're sitting in 61 degrees never known a February like it.

    Here, it's 30 outside, but 79 inside. That's thanks to the gorilla sized output of my wood stove. (My wife loves it but it forces me to wear shorts at times.)


    Don't buy into the "global warming" thing. After I retired, suspecting that the climate topic might be something along the lines of Y2K and the Myan Calendar, I researched the data. (Trust no one - look at the data.) I found that temperture data has been manipulated, actually skewed upward. There are reasons given for adjusting the numbers, but they have nothing to do with science. It's about continuing research, and "funding". (Truly sad.)


    What most people are not aware of are the long standing cycles, from a paleoclimatology perspective. Right now, believe it or not, we're living in an ice-age and we're right in the middle of a warm cycle that lasts, very roughly speaking, 14 thousand years or so. (All of recorded history is in this cycle.) In the cycle averages, the planet could be fully 5 degrees hotter than it is right now and all would be absolutely normal. (We're not talking about tenths, to one degree.) In fact a paleoclimatologist commented about this cycle being quite "mild", as in being a bit cooler than usual. In the bottom line, one human life span is nowhere near long enough to make meaningful observations on climate. It's been much hotter and colder, before man kind had any influence.
    If you're interested, the CO2 thing is bogus as well. There's griping about 350 ppm being too high, yet there are ice cores from Antarctica that shows 6000 ppm during periods when it we had "Ice Ball" Earth. Climate change is a reality that will happen, as it has for literally millions of years.
    If we were to "invest" money in anything, my preference would be for decent weather forecasting that has better than "flip of the coin" accuracy more than two days out. (That's something I could use.)


    Well, power is back! Time to get to work. (Sorry 'bout the meandering. :) )

    • Offizieller Beitrag

    (Sorry 'bout the meandering. )

    No that's the idea, perhaps we should change the title "An Evening with Dr Flmaxey" :D
    _________________________________________________________________________


    On a different note I see you have added a reference to Watchtower in the Pi-Hole howto :thumbup: I started investigating this further and for the life of me I can't get my head around it, there does appear to be a way to run Watchtower without updating Pi-Hole.
    I've found some reference to this on github.


    How to disable watchtower on a single contiainer
    Exclude container #222


    This is from the github;


    By default, watchtower will watch all containers. However, sometimes only some containers should be updated.
    If you need to exclude some containers, set the com.centurylinklabs.watchtower.enable label to false.
    LABEL com.centurylinklabs.watchtower.enable="false"


    Or, it can be specified as part of the docker run command line:


    docker run -d --label=com.centurylinklabs.watchtower.enable=false someimage


    If you need to only include only some containers, pass the --label-enable flag on startup and set the com.centurylinklabs.watchtower.enable label with a value of true for the containers you want to watch.
    LABEL com.centurylinklabs.watchtower.enable="true"Or, it can be specified as part of the docker run command line:


    docker run -d --label=com.centurylinklabs.watchtower.enable=true someimage


    Reading through the above the command line entry would be --label-enable --label=com.centurylinklabs.watchtower.enable=false pi-hole (i.e. pi-hole would be depend on how the user had specified the name) --cleanup (I use cleanup and I find it works)


    That would prevent pi-hole being updated by Watchtower, if I have worked this out correctly.


    However, it does not state if any of the command line entries require a , to separate them, I'm assuming they don't.

    • Offizieller Beitrag

    I discovered an issue in the set up or configuration of unbound; when unbound is directly installed on an OMV host, AND pi-hole is running in a Docker. This combination OMV+unbound, with pi-hole in a MacVlan'ed Docker works, but it has an inexplicable side effect.


    The OMV host itself, can't resolve external names. It returns, "Temporary Failure to Resolve" and I've seen this error on two different hosts so far. This issue, unfortunately, prevents OMV from updating. Adding to that, the Docker container (pi-hole) doesn't respond to "ping" from the OMV command even using the IP address. Further adding to the oddness is that all name lookups work fine for LAN clients, it's just OMV host itself that has issues. (I tried the normal stuff, setting OMV's DNS address to 127.0.0.1, OMV's IP, the router, and others.)


    I haven't looked at work arounds, like stopping and restarting unbound service, fiddling with ports, etc., so I pulled the unbound section of the "How-To".
    Despite the "intermediate user" threshold warning, beginners were trying to set it up, failing, and asking what went wrong.


    I'm increasingly of the opinion that the "dedicated appliance" approach to running pi-hole with unbound is the way to go.
    __________________________________________________________


    WatchTower - I don't get it and see no use for it. A docker container = a VM = a full, if barebones, host. Watchtower is not unlike, allowing M$ to auto-upgrade a Windows client. As it was with the Win10 auto-upgrade, broken install's are inevitable. It seems to be an unnecessary risk but, to each their own.


    Edit: I looked at your links, RE disabling WatchTower on a "per container" basis. I have yet to use Watchtower, but the process seems protracted enough to where I can't see beginners getting that consistently right. (Flip a coin.)

    • Offizieller Beitrag

    I discovered an issue in the set up or configuration of unbound; when unbound is directly installed on an OMV host, AND pi-hole is running in a Docker. This combination OMV+unbound, with pi-hole in a MacVlan'ed Docker works, but it has an inexplicable side effect.

    Is this with the latest version of Docker, when I was running this in Docker and using unbound it was working, so have they made some change that would cause problems using unbound? I would assume that a standard Docker install of Pi-Hole would be Ok?


    Re; Watchtower I appreciate what you're saying, I use it but I don't have many Dockers running but it saves me updating each one and I'm sure that is why most people use it, I was just wondering if you would cast your eye over what I have posted, particularly in relation to the command line to prevent Watchtower updating Pi-Hole. My conclusion seems to make sense from the links and some further reading, but with some users having a number of Dockers running one of which is Pi-Hole, Watchtower can mess up that container.

    • Offizieller Beitrag

    RE unbound:
    This issue is present with latest and the previous version of pi-hole in Docker. It could be an unbound config issue (maybe go with port 53 itself?), it might have something with a pi-hole/Docker config parameter. I really don't know. The behavior is odd to say the least and it doesn't to seem the have any effect where pi-hole and unbound are installed on the same host. (The Diet-Pi for example.) I suspect a script work around for updating OMV is possible, but the OMV/pihole-in-Docker/unbound "combo" is not a topic for a beginner "How-To".


    RE Watchtower:
    There's no right or wrong here, it's just a matter of preference. You know my take on it and I understand yours.

    I was just wondering if you would cast your eye over what I have posted, particularly in relation to the command line to prevent Watchtower updating Pi-Hole.

    I did, and I followed the links to the errata. (It looks like the change was added to the code; com.centurylinklabs.watchtower.enable="false") But without WatchTower being installed, and never having used it, there's no context. If you want to write a How To, I'll try to help, but my reservations lay with the command line and beginners. Keeping things in the GUI, seems to get the best results.
    I'm giving thought to config'ing up a hardware platform for some BTRFS torture tests, just to see what it can take. This might be a good platform to test Watchtower, on real hardware, without polluting up my servers.
    __________________________________


    Oh, yeah, since you're interested in SNAPRAID and UnionFS:
    I setup both, using older hard drives, to see if I could get some real world results in less than a few years. I ran a "fix silent" and 8 bit errors were corrected, on clean drives and with clean EXT4 file systems. Of additional interest is that the files, with corrected errors, are listed by name. Awesome! And while I haven't run a sector editor test for independent confirmation, I believe that the advertised bit-rot protection and correction are real.

    • Offizieller Beitrag

    The OMV host itself, can't resolve external names. It returns, "Temporary Failure to Resolve"

    I've just been re reading your posts regarding OMV+Unbound+Pi-Hole, I've also found my old notes when I was using this and I certainly never had an issue with OMV updating.


    My OMV is set as a static IP with a DNS entry of 1.1.1.1, Pi-Hole Docker was using the MacVlan as per your howto, following this and making the necessary changes initially it didn't work and it had something to do with the fact that I was not using Pi-Hole as a DHCP server I was using the plugin, this was because certain network hardware would not communicate with Pi-Holes DHCP as it was looking for a hostname before assigning an IP.


    So I did two things (and that was after some exchanges on here) the Custom 1 (IPv4) in the DNS settings I changed to the IP of my OMV server but I also changed the port to 53 in the /etc/unbound/unbound.conf.d/pi-hole.conf, and therefore in the Custom 1 DNS setting


    So for the above statement to be true "Temporary Failure to Resolve" it would mean that OMV would be using Pi-Hole as a DNS, or am I coming at this totally from the wrong direction.
    _____________________________________________________________________________________________


    Thanks for the Watchtower confirmation, all I was thinking of was to post something within the Docker section in respect of Pi-Hole not using Watchtower to update, having read a post where someone was using Watchtower and it updated Pi-Hole. Watchtower seems to be useful if you are using things like Sonaar, Radar, Transmission, Emby etc. it works and works well, but with the Pi-Hole devs their updates usually require a complete removal and reinstall especially when using Docker. Whereas on dedicated hardware pihole -up is sufficient to run and update Pi-Hole.

    • Offizieller Beitrag

    Replaced router :thumbup: this means I can set Pi-Hole as the default DNS server and therefore disable the OMV DNS/DHCP plugin, this router also runs IPv6 as well as IPv4, but the IPv6 cannot be disabled this can only be done on the client.

    • Offizieller Beitrag

    My OMV is set as a static IP with a DNS entry of 1.1.1.1, Pi-Hole Docker was using the MacVlan as per your howto, following this and making the necessary changes initially it didn't work and it had something to do with the fact that I was not using Pi-Hole as a DHCP server I was using the plugin, this was because certain network hardware would not communicate with Pi-Holes DHCP as it was looking for a hostname before assigning an IP.

    Really..? After fooling around with the error issue, on two separate hosts, that is interesting.
    And I didn't know pi-hole (DHCP) looks for a hostname, before spitting out an address. But as I remember it, pi-hole allows for mapping a name to a reserved DHCP address. (Kind of important, in my case, for IP phones.)


    So I did two things (and that was after some exchanges on here) the Custom 1 (IPv4) in the DNS settings I changed to the IP of my OMV server but I also changed the port to 53 in the /etc/unbound/unbound.conf.d/pi-hole.conf, and therefore in the Custom 1 DNS setting

    I think you eliminated the issue before it happened with, really, common sense. The custom port didn't sense at the time (5353), but I was focused on getting OMV and the Docker to talk so I didn't question it and, since I moved to the DietPi, I didn't notice the side effect at the time. I'll check that out - thanks.


    But I still have reservations about re-posting the unbound add-on. While I don't mind, absolute beginners contact me out of the blue, about why something doesn't work. I'm thinking of that RAID thread, where the user pulled the wrong drive AFTER being specifically told to look at the drive's info (serial #, etc.) in the GUI, so he'd pull the right drive. :)


    After posting a How-To, that has been combed through eliminate ambiguity, some users will change the process then they contact me when it doesn't work. That recent pi-hole in ESXi thread comes to mind. (With all of his plugin's, I imagine ryecoaaron deals with this all the time.)


    So, while I'm not trying to discourage you from posting a WatchTower How-To, fix, etc. (I'm sure it would help the majority) there's a down side to it. Support and maintenance. The old adage applies; "No good deed goes unpunished."
    ________________________________________________________________________


    OMG - I just concluded two extended encounters with 60 Grit. I know I'm preaching to the chior here but:
    I've never seen one who becomes so fixated on the low level stuff (the minutia) that the big picture becomes abstract, almost non-existant. And the crux of it all (both events) was trying to prove me wrong, that BTRFS doesn't need an UPS, for a specific reason I stated. The rest was merely add-no nonsense, swirled in, to support his case. (Because, as you know, he's gotta' be right. :rolleyes: ) The really interesting aspect of it all is, where BTRFS is concerned, he's merely speculating based on things he's read. He hasn't tested it as one could not possibly test all scenarios. And while he might be right, truly, it's irrelevant in either case.


    In both cases, by his own admission, with all of the pontification and attempts to educate the "rest of us", none of it had anything to do with the topic at hand. Fact is truly stranger than fiction. I suppose I could never author a novel like "One Flew over the Cuckoo's Nest" because I couldn't dream this stuff up. :D


    And the little serialistic comedy is still on-going in the proposal. Despite being called out directly, "give others a chance to freely comment without pot shotting them", he's still doing it! :) (Today, in fact.) A user commented, his opinion - he's entitled to it, where 60 grit immediately proceeded to educate him on why users need not be concerned about the file system they're using. I'd considered muting the the tread but now it's about a morbid sense of curiosity, to see him morph from; I'm **not** participating in the whole 'Should OMV use btrfs as default fs?' discussion **at all, to telling users what they need. You know, "the details of dealing with file systems are just too hard for them". This stuff is better than "comedy night", at a bar.

    • Offizieller Beitrag

    And I didn't know pi-hole (DHCP) looks for a hostname, before spitting out an address. But as I remember it, pi-hole allows for mapping a name to a reserved DHCP address. (Kind of important, in my case, for IP phones.)

    The hostname was the only thing I could come up with in the OMV plugin lease would have (*<ip address assigned>) as the hostname, even when I attempted to add it (after setting it up first on the box) as a static IP with the mac address then giving it a hostname purely for pi-hole's DHCP it still would not accept it. It seemed an odd behaviour of pi-hole's DHCP, this replacement router on the other hand assigns a set of 8 letters in groups of 2 if the machine does not return a hostname, which is fine because you can edit that to assign a name to the ip.


    But I still have reservations about re-posting the unbound add-on.

    I know what you mean you add in bold capitals "Warning follow these instructions in the order below" someone will still not get it working.
    My Watchtower thought was not to post a how-to per se, but to highlight the Command option under the Extra Arguments, so that Watchtower could be used but Pi-Hole is removed from the auto update, and other command options.


    As to fs I try not to get involved in discussions regarding file systems, I rely on what I know or what I can find out, I prefer the 'tried and tested' approach. It's like not using Raid for home use TBH if someone has the knowledge and runs a backup then let them use it, I currently do but that is only because I have an understanding of it, so what if it's not necessary for soho. But I still have sights on moving to SnapRaid and UnionFS :)

    • Offizieller Beitrag

    I rely on what I know or what I can find out, I prefer the 'tried and tested' approach.

    I'm of the same mind set. I'm not in the school of thought where all change is good. Universally speaking, it's not. Usually, one is trading one set of problems for another. The only reason why I use ZFS was to address the bit-rot question and to prevent corruption from slowly failing drives. But, even with it's long and venerable history, ZFS was adopted only after extended testing. The key is "backup". One can take risks with backup. With a few levels of backup, one can safely take on more risk with relative safety.


    And that's what is key to the issue and the overall question at hand:
    If you know the majority (or even a large percentage) of users do not have backup, why would you force the use of single, unproven file system that has so many known caveats? In any case, this an open ended topic. Nothing is set in stone, yet, and I know I can roll my own server.


    But I still have sights on moving to SnapRaid and UnionFS

    And I think that's a good thing. In "bang for the buck", as in the least drive real-estate required for an extensive list of benefits, there's nothing that I know of that's better than SNAPRAID.


    Can use stable file systems with well developed maintenance tools, check. Restore silent corruption (bit-rot), check. Restore files and folders, check. Replace and restore a drive, check. If you throw rsnapshot in there, you also have snapshot like (versioned) backup on sensitive shares as well. In the sum total, that's most of the benefits of a CoW file system without the down sides. Perfect for home use.

Jetzt mitmachen!

Sie haben noch kein Benutzerkonto auf unserer Seite? Registrieren Sie sich kostenlos und nehmen Sie an unserer Community teil!