The Class E Network

    • Offizieller Beitrag

    I'm afraid SWMBO has already past sentence and the appeal date is not til next year :D


    I think you handled the ZFS with aplomb, but there were two choices, say nothing, or reply with a simple straightforward viewpoint, you choice the later, he was the one that felt a reply was necessary.


    Re the forum pontiff; that would have been something, could have gone on for pages ^^


    Does that mean I should wait for your testing scenario for UnionFS+Snapraid :) cos you're now going to get a dumb question -> does that mean in the event of say a drive failure you deal with it through Snapraid first i.e. remove said failed drive from Snapraid, then remove from UnionFS.


    You've never heard of SWMBO I think it was used in a comedy sketch but it's always stuck, but I only ever it in forums :) It's like my phone she's not listed by her name, she's listed as 'Err Indoors' but again that's British slang from the TV series Minder, and her ringtone is Fawlty Towers :D:D I think both are featured on You Tube.

    • Offizieller Beitrag

    Does that mean I should wait for your testing scenario for UnionFS+Snapraid cos you're now going to get a dumb question -> does that mean in the event of say a drive failure you deal with it through Snapraid first i.e. remove said failed drive from Snapraid, then remove from UnionFS.

    No, I'm thinking SNAPRAID is fine but there's a difference between thinking and knowing.


    There are a few failure recovery modes I really want to test.


    1. The SNAPRAID drive dies. Seems simple. I think you'd have to remove the content files and other markers from data drives, get another parity drive, sync. (But is it that easy? :) I'd have to look at the doc's)
    2. Data Drive Failure
    This is where UnionFS would be involved. I imagine you'd have to replace the drive with something that could hold the previous contents, but is still smaller than the parity drive. Then it would be, add it to the Union, and restore with SNAPRAID. (I'd want to do a walk through recovery.)
    3. Selective file/folder restore (Delete something a do a restore.)
    4. Corruption detection and correction. (Very slight file corruption is easy with a sector editor. Corruption detection and correction is what I'm most interested in.)


    When I get this Intel box outfitted the way I want, I'll be VM'ing my butt off. Realistic testing will be easier. While VM's are not perfect simulations, they're close enough for this purpose.

    • Offizieller Beitrag

    I've been briefly reading the Snapraid manual 8o I had to sit down with a cup of sweet tea -> for the shock :D


    The Snapraid option appears to fairly straightforward regrading drive replacement, whether it be the parity drive or a data drive, the only caveat that would concern me is that it suggests that the 'array' is not accessible whilst doing a recovery -> this is in the manual "The first thing you have to do is to avoid further changes at your disk array. Disable any remote connection to it, any scheduled process, including any scheduled SnapRAID nightly sync or scrub."


    That may or may not be a problem for me, on Emby I use a Kodi sync plugin (I find that easier to use than the other option) that sync's your shares from the server to Kodi, not sure if I would have to shutdown Emby and the Pi running libreelec.


    Recovering/Reconfigure also allows you to use an external usb drive, could be useful.


    I think what is confusing me is if you have a drive failure that is part of your UnionFS pool I get the impression that you recover using Snapraid before adding the drive back to the pool -> why? because Snapraid will recover the data from the parity drive, adding a drive back to pool empty would create a problem. At least that's how my brain is working.


    UPDATE: I can now report that the DietPi PiHole is acting as the main DNS :thumbup:

    • Offizieller Beitrag

    the only caveat that would concern me is that it suggests that the 'array' is not accessible whilst doing a recovery -> this is in the manual "The first thing you have to do is to avoid further changes at your disk array. Disable any remote connection to it, any scheduled process, including any scheduled SnapRAID nightly sync or scrub."

    This is another reason why a sim'ed recovery is not a bad idea. What's involved, for ones own use case? In a recovery, attention tends to be on the task at hand (even beginners), but details are often missed.


    The R-PI (pihole/unbound) is doing fine here as well so (there) I just pointed my router to it. I'll do the DHCP change later.
    _____________________________________________________________


    The adaptec 5405Z controller that came with the server is driving me nuts. It doesn't pass full SMART data for drives attached. It reports that an attached drive supports SMART, that the drive is OK (whatever that means), and a temp of 0-C. It was two firmware builds out of date so I flashed. No change.


    What amazes me is, I know this controller is still in use out there, because it's not a bad unit for this purpose where external expanders are unlikely. (It has one 6 drive internal expander that uses 2 ports from the controller). What bugs me is, this controller was in use (supported), maybe even sold, until 2010 at least. At that point SMART was mature, with full support for reporting attributes . I can't find anything on this controller, not providing full smart attributes to the host. Nothing - not even an unfounded gripe of some kind!
    __________________________________________________________________
    Take a look at this smartcrl output:


    smartctl --all /dev/sda
    smartctl 6.6 2016-05-31 r4324 [x86_64-linux-4.18.0-0.bpo.1-amd64] (local build)
    Copyright (C) 2002-16, Bruce Allen, Christian Franke, http://www.smartmontools.org


    === START OF INFORMATION SECTION ===
    Vendor: WDC
    Product: WD20EZRZ-00Z
    Revision: 80.0
    User Capacity: 1,997,149,306,880 bytes [1.99 TB]
    Logical block size: 512 bytes
    Serial number: WD-WCC4M2VT6CHD
    Device type: disk
    Transport protocol: SAS (SPL-3)<- It shouldn't matter, but it's a sata drive.
    Local Time is: Sat Nov 24 10:37:42 2018 EST
    SMART support is: Available - device has SMART capability.
    SMART support is: Enabled
    Temperature Warning: Disabled or Not Supported



    === START OF READ SMART DATA SECTION ===
    SMART Health Status: OK <- Without attributes, how is this determined?
    Current Drive Temperature: 0 C <- These lines are pure BS.
    Drive Trip Temperature: 0 C


    Error Counter logging not supported


    Device does not support Self Test logging
    ______________________________________________________________________


    Crappers - ryecoarron told me that this is common with RAID controllers. That's a bummer. I liked getting E-mail notifications if, for example, an unallocated sector popped up. Maybe I'm being too cautious. A ZFS scrub would (probably) show a few errors before a drive fails.


    Oh well, the perc H200 arrived yesterday. It's faster and with the IT firmware flashed onto it, I might be able to get SMART data.
    ________________________________________________________


    EDIT - sorted! (sort of):
    Since I had the command line agent installed, I'm using a couple command lines to query the controller and pull drive SMART stat's, in the GUI's scheduled tasks. At least this is workable.
    Now, to figure out a script for notification if something goes south and I'm trying to get a remote connection to the controller as well.

    • Offizieller Beitrag

    Bye, Bye, Adaptec 5405Z - hello Perc H200.


    Man, I was thinking I might have flushed a whole $26 USD's down the drain but I finally got the Perc flashed to IT mode. ($26, what a deal, but it took weeks to get here.)


    No more delayed boots, while the adapter cycles through BIOS, full smart stat's, and enough physical ports remaining (not including the 6 drive bay) to connect 9 more drives. I think I'm good. :)


    Now for rdimms and another CPU.

    • Offizieller Beitrag

    The Adaptec's need the installation of the Storage Manager Software to be installed to get any information from the onboard Adaptec bios and drives attached. My last foray with these wouldn't allow a drive to be added as a hot spare, it would 'see' the attached drive but that was it. I had to wait for the next Software update for it to work!


    Well at least you managed to flash the Perc, that's a lotta drives :)

    • Offizieller Beitrag

    So, I have to ask, what do you think about the current Brexit deal? As the "deal" is being portrayed here (from the fake news outlets) Britain would still be, functionally, in the UN. As I understand the proposed deal, "Brexit" would amount to being able to take control of the borders again, but little else. With news propaganda being what it is, I imagine that my understanding is w-a-y off base. You'd have to have a much better understanding. (The Sun must be headlining this topic, along with the page 3 girl.) Do tell. :)


    ____________________________________________________________


    (Even after Flashing the Adaptec to latest firmware build.)

    The Adaptec's need the installation of the Storage Manager Software


    I had ASM installed, on the command line. The adapter's summary info dump was still similar, where generic drive information was concerned, to what I posted above.


    The commands I used were:
    ./usr/StorMan/arcconf GETCONFIG 1 (Dumps the status/health of the controller and hard drives.)
    ./usr/StorMan/arcconf GETSMARTSTATS 1 (Dumps the supported attributes of attached drives.)


    The latter command provided all drive supported attributes but, get this, the attribute "numbers" were in hex without names.
    To make it a bit more friendly, I installed ASM for windows (graphical) for a remote connection from a client. While the required agent was running on the server, I couldn't get it to connect in the first attempts.


    With the Perc laying there, these issues just didn't seem worth running down and I'd already had it with compromises that seemed to come with using the Adaptec. It was just the matter of flashing the H200 to IT mode (another minor adventure) - making it a JBOD controller with transparent SMART pass-through.
    ________________________________________________________


    With SMART data passed through to the host (as opposed to being to forced to monitor it from time to time) E-mail notifications can be set up in OMV. I know most don't even bother with SMART stat's but, from my view point, SMART stat's are the only indicator available to look at the weakest components in a server. (Even if it is something of a Crsytal Ball.)


    In any case, while I doubt I'd stress the Adaptec's 3GB/s per port, the Perc H200 is a newer 6GB/s item.

    • Offizieller Beitrag

    Just a quickie, in the middle of cooking dinner (supper) this is link to adaptec re your card to get the information you want requires the Storage Manager -> but!! I fail to see how this would work on a headless system, I've only used it on MS systems, the card in essence is useless without the Storage Manager Software, as reads the information passed from the cards bios, so you can obtain all the smart data, warnings of any potential drive failure all of which can be sent via email -> but it can only be done with that software.

    • Offizieller Beitrag

    You know, back in the day, good SCSI RAID controllers were uber high end items. They supported remote configuration connections, even during the days of Windows NT. I informed the security folks back then, that we needed to secure those controllers with at least a minimum of a unique password. (As opposed to leaving them at the default which could, even back then, could be looked up on-line)
    When I remoted into one and the menu came up; I asked, what do you thing would happen if I took the array off line and initialized the drives? Taking about a crushing denial of service attack!, and restoring from tape... Needless to say, they got on it with a quick policy change.
    _________________________________________________


    You know, it's almost shocking how hot these modern RAID controllers are. The Adaptec had an unstressed running temp of 80C, and the controller dump had this labeled as "NORMAL" :!:
    Since I suspect the same would be true of the Perc H200, I think I'm going to install that mini fan I bought for the EX470 on the Perc's heat sink.

    • Offizieller Beitrag

    As I understand the proposed deal, "Brexit" would amount to being able to take control of the borders again, but little else.

    Yep that's about it, but the BBC website is the best place to get an update, tbh, if we're going to leave then leave, what we are going to get is a half in, half out, makes no sense.
    David Cameron had the best option, stay in and change the EU from the inside, there's a lot of it I don't understand myself, the leave campaign had no credible strategy to leave, they still don't! But the current proposal is a farce, but the problem is the UK has been integrated for so long we need each other, but the EU aren't going to make this easy, the UK is the first but it might not be the last!


    The next problem is Italy, their budget has been rejected by the EU and have been to told to resubmit!


    I voted to stay simply because leaving would create numerous problems, but it would give the following generations an opportunity to make changes. The next 20 years, maybe 10 will give an idea on how the UK is fairing, unless there is another referendum :rolleyes:

    • Offizieller Beitrag

    Well, from what I know if it from the financial end alone, the EU couldn't possibly work over the long term "if" some countries are allowed to over-spend, like drunken sailors, while others (England, Germany) foot the bill. Greece is at the top of the list, followed by others with Zero spending discipline. A common currency (the Euro) might work if the buy-in was like casino chips, meaning individual country currencies should float on the market against the Euro. In that way, the over-spenders could trash out their own economies without affecting the whole.


    But what do I know? I do know how to balance a check book and I'm legally obligated (by my government) to do so. The obvious question to ask would be, why are governments NOT obligated to do the same?
    ________________________________________________________________


    It will be interesting to see what the future holds on both side of the pond, but I hope it won't be "interesting" in the sense of the notable curse; "may you live in interesting times".

    • Offizieller Beitrag

    I bought a "pair" of Xeon's for my mobo, on Ebay,. for $5 delivered. :?::!: There must be a lot of servers, being decommissioned in this generation. The extra heatsink and fan combo is going to cost way more than the processor. I'm bidding on the spec'ed Intel heatsink/fan combo from the doc's, but I'll go after-market if I don't get it. I also have 32GB (4X8GB) in rdimm's on the way.
    With these upgrades, I'll be ready for ESXi on the cheap. :)


    There's some really good deals out there, but, it's a matter knowing the hardware and what to look at. For example this Intel server is a bargain. The Dell equivalent, in nearly all hardware spec's, is significantly more expensive on the used market. While that's an outcome of supply and demand, there's really very little to account for the difference.
    ________________________________________________


    RE Brexit:
    Finger pointing,,, that's guaranteed to happen in any case. (It's the adult equivalent of "blowing a raspberry".) Tiresome isn't it? As if it helps, somehow, to remind others of the hypothetical reasons why something didn't work. That's very much like history. Humanities' steadfast refusal to learn from history often makes me wonder why we continue to record it.


    we both need each other whether both sides like it or not :)

    No doubt and I completely get the reason behind the EU, as it was originally conceived. Much as it still is in the middle east, in times past, wars in Europe were a long standing pattern. That needed to stop. But the EU, as it is today, is very different and covers many more aspects of international affairs, commerce, etc., when compared to a simple non-aggression treaty.
    The question is, what to do when one or the other misbehaves? The concept of "Sovereignty" allows some EU members to stamp their foot, when something comes up that they don't like (such as the need for a reasonable budget), and do whatever they like. That makes things "interesting".


    And while many choose to ignore it, being decidedly around the 20 trillion dollar debt corner, we're in no better shape over here. (At 20 to 22 trillion in debt, many financial experts agree, there's no return.) So, on this side of the pond, we're headed toward "interesting" times as well. I'm just hoping, selfishly, that the reckoning, (paying the piper, etc.), doesn't happen in my lifetime.
    Such nonsense would interfere with my hobbies. :)

    • Offizieller Beitrag

    I need a favor if you would:
    Check out the Win 10 "How-To" and see if the steps make sense. Since I can't lay my hands on my W10 box to walk through it, I wrote the addition in the blind.


    Also, I installed ASM on my Intel Box and since the ASM package was way out of date, I had to compile a DEB package to do it. While jumping through the hoops to install it is one thing, removing is another. I was thinking about rebuilding to get rid of it, to have a clean install, or since the package and dependencies don't take much room, should I just ignore it? What do you think?

    • Offizieller Beitrag

    Windows 10:


    Your How-To is fine -> IMHO you're jumping through too many hoops. (Apologies if some of this is not in any form of logical order)


    A lot of problems started from the Creators Update, biggest PITA ever, created so many problems for end users, but that's the norm for MS, subsequent updates appear to have resolved the initial issues. So the registry options should not be necessary and the current 'pinned' solution, technically, is out of date. Anyone who has a new W10 machine should not need to do any of that, likewise W10 Creator problems have been resolved so none of those options are relevant.


    I have 6 machines running W10, 2 running 64 bit and 4 running 32 bit (two of which are netbooks!) The 64 bit versions have been the ones that have caused the problems, for whatever reason in the early days updates would error. Now, both are running 1803 and behaving, the 32 bit versions are the W7 updates, never been a problem.


    All of my shares are guest access so my observation to Molok's solution is a mystery to me.


    I believe Molok's fix came about because he completed the reset/recovery option, something I would never do, nor advise, do a clean install, in essence it's the same thing except with the clean install you need a backup of your working files. Something the average end user probably never does, so they opt for reset/recovery. I have completed a clean install on my W10 64 bit laptop twice due to update issues and I have never had to set that Lanman option -> technically a clean install should set up W10 the same as reset/recovery.


    I found my own solution from the MS help forum and it worked! IMHO that is far simpler for an end user than running gpedit, then if doesn't work you have to do gpedit /force /boot because invariably the computer will need to restart to enable the change.


    The BIG problem which MS have failed to resolve from the get go is name resolution, it does once the computer has been on the network for x amount of time, then you can resolve hostname to ip. The issue with W10 is MS networking relies on Function Discovery Resource Publication and WS-Discovery, so initially it will not locate something like OMV by hostname only by IP, therefore it will not display in File Explorer -> Network -> never ever!


    The MS solution now once your nas is located is to map network drives, that way they always accessible unless your nas goes down.
    _____________________________________________________________________________________


    RE ASM; I've installed stuff before, then decide not to to use it :rolleyes: only to leave it on the system 'because it's not doing any harm' Your only other option is remove --purge to remove it.
    ______________________________________________________________________________________


    Learn't something today, running the DHCP plugin for a while on OMV and I have removed the second Pi-Hole from the DNS settings so it's no only picking up DietPi. The wife is at home this morning (see SWMBO is quicker to type than The wife) :) -> she has no internet access ?(
    Well that's like signing your own execution order :D turns out the ip address has been renewed according to OMV but it had failed to issue (for whatever reason) the DNS of DietPi, so a simple /release /renew solved it, just seemed odd everything else on the network has been fine.

    • Offizieller Beitrag

    The one thing about the US you can get PC equipment cheaper than here! It's things like the Odroid and RockPro64 they are based in the US or at least you can only purchase from the US, looking for those boards in the UK they are double +
    If I purchased the RockPro64 I would have to pay import VAT (there's no customs duty as it's under £135) then there would be shipping costs, then there could be another charge of up to £8 for collecting the import VAT depending upon the shipper.
    ________________________________________________________________________________


    Looks as if Brexit is not dead, Trump now saying trade deal may not be possible! Group of MP's have gone to the European Court of Justice to examine whether the UK can call off the process of leaving the EU without permission from the other member states.


    What's the saying "It's not over til the fat lady sings" I often wonder where I would be if I had taken the job at Disney Florida when I was 20.

    • Offizieller Beitrag

    Back again, just noticed your post here re docker have been having problems with mine and I am contemplating starting over due to some errors but my output is


    But this is my Kernel Debian GNU/Linux, with Linux 4.9.0-0.bpo.6-amd64 never updated it.

    • Offizieller Beitrag

    On ASM: I made a DEB package from a tarball and installed adependency that I can't remember the name of it, so I have no idea ofwhat, exactly would be purged and what it might take with it. For other reasons and wanting a clean build, I'm leaning towardrebuilding anyway. With a clean 3.0.X build on the TS140, andin being highly selective about updates and add-on's, I've had zeroissues with it.


    Zitat von geaves

    Looks as if Brexit is not dead, Trump now saying tradedeal may not be possible!


    Just about anything Trump says is a position (a posture) inanticipation of a future negotiation. "The deal" iswhat makes him tick - he wrote a book about it. (And it wouldbe wise of anyone who deals with him to read it.) There will bea deal but, if it's a good one that benefits all who sign it, no one will bethrilled with the outcome. That's as it should be.


    Zitat von geaves

    Group of MP's have gone to the European Court of Justiceto examine whether the UK can call off the process of leaving the EUwithout permission from the other member states.


    [/quote]
    If I was you, whether you support leaving the EU or not, I'd be moreconcerned about a Government that feels empowered enough to ignorethe will of the people, after putting an issue on a referendum. Since nothing is 100% certain, why bother to add the issue to aballot, if Government can choose to ignore the outcome? Andwhat's their counter argument for not following through; you "thelittle people" don't know what's good for you?
    There's anold axiom from military leadership; never give a command that youknow won't be obeyed. If there was any chance,whatsoever, that the Government would not execute the outcome of thereferendum, in good faith, one way or the other, the issue should never have made it to theballot. This one issue and its' handling, in and of itself, ismore than enough reason for May, her predecessor, and several more that were involved, to resign.
    (That's just an opinion. Don't do a "Jody" on me. :) )

    • Offizieller Beitrag

    On ASM: I made a DEB package from a tarball and installed a dependency that I can't remember the name of it, so I have no idea of what, exactly would be purged and what it might take with it. For other reasons and wanting a clean build, I'm leaning toward rebuilding anyway. With a clean 3.0.X build on the TS140, and in being highly selective about updates and add-on's, I've had zero issues with it.
    ________________________________________________________________________________


    On Win10:

    Your How-To is fine -> IMHO you're jumping through too many hoops.

    I'd agree but there's so many damxed installs out there, in so many different states, it's impossible to nail down a single process or even a small set.

    All of my shares are guest access so my observation to Molok's solution is a mystery to me.


    I believe Molok's fix came about because he completed the reset/recovery option, something I would never do, nor advise, do a clean install, in essence it's the same thing except with the clean install you need a backup of your working files.

    That's one of the over arching problems with Windows 10 (and Windows in general):
    Some users do things that are not best practice, Win10 let's them do it and while the install continues to work, it's in an odd ball state.
    ((And while it may be best practice, how many users do you think would do a clean install, if they think they can avoid it?))


    Given what's in the Win10 thread, at least 2 users experienced the same problem where the security policy edit "fixed" it (so to speak). With the number of OMV users out there, that means at least 20 OMV users (idle speculation, more?) may have to deal with the issue. Since their Windows install "works", and sorting out errata on M$ issues is nearly impossible, they're going to come here if the problem is related to OMV.

    So the registry options should not be necessary and the current 'pinned' solution, technically, is out of date. Anyone who has a new W10 machine should not need to do any of that, likewise W10 Creator problems have been resolved so none of those options are relevant.

    Each of the hoops, in that oversized how-to, addressed at least one potential issue that blocked access to SMB shares, at some point in time. Since some users turn off updates or only allow approved updates, and there's business PC's in closed networks to consider, there are innumerable PC's out their that still have old connectivity issues. (And I'll admit to thinking about turning off automated updates myself. I get tired of having my PC hijacked for 10 minutes or more when I want to use it.)

    The issue with W10 is MS networking relies on Function Discovery Resource Publication and WS-Discovery, so initially it will not locate something like OMV by hostname only by IP, therefore it will not display in File Explorer -> Network -> never ever!

    I did a bit of research into Win10 network discovery and found, as I remember it, 6 services where at least part of their function is network discovery. (And I set all of them to "Start" not "Auto".)
    In any case, setting a hosts file entry should completely negate the host name to IP address mapping issue, if the OS utilized it as the override/user defined default it was originally intended to be. At the client level, for local DNS, the hosts file still works. Otherwise, it appears that the OS ignores it. (And it can be a PITA to set an entry in hosts. It can require Admin access in Safe mode - ridiculous for a file that only needs to be read on start up and is, otherwise, ignored.)
    If nothing else does it, the hosts entry tends to work for mapping drives and creating short cuts, but there's still nothing in File Explorer -> Network.


    _________________________________________________________________________________________


    I have one 64bit W10 box which isn't here at the moment. Given issues that are cropping up with the wife's (Vista) laptop, and considering the user :) , I'm thinking about testing a W10-32bit install on it. After the build, I'd have a clean install to see how things work out of the box, with a current W10 build. At least, I'll see if the components of the original problem(s) still exist which were, "can't map a network drive" which means short cuts won't work either and, in some cases, "can't connect to SMB2 shares". If these problems still exist, what it takes to get W10 on line.


    Matching workgroup names (OMV to W10) is a given, along with insuring that W10 network discovery is on.
    As for the rest of the How-To (given your experience), maybe you'd consider ranking the remaining "stuff" and it's chance of fixing connectivity, from "mostly likely" to "least likely"?

Jetzt mitmachen!

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