Posts by HannesJo

    I have a simillar "problem" with my setup. I have two backup disks scheduled for weekly rsync tasks, and both are under samba shares. I am connecting these disks manually to my OMV machine only for backup purposes.


    Because of that I have a lot Syslog and Demon logs - 36 entries in one minute.


    Maybe there is some safe wat to unmount this kind of disk in the system without disabling all sharing points?


    No there isn't. The usage of temporarily added drives for services is very uncommon. This is simply not how a server is supposed to work. So if you insist in doing it this way, you gonna have to disable shares and unmount manually. Of course you could just program it yourself but this is not going to be easy.


    But again, error messages are not a problem. A problem is misusage by disconnecting mounted drives. The error messages is just your server telling you that sth bad happened.

    In general, when you expose ssh to the internet it is very common someone will find it and try to connect. It is an automated process where people search for vulnerable devices in the internet. If you use PublicKey authentication, you should actually be quite safe. You could also setup a fail2ban service that bans an ip after certain number of failed login attempts. It is also a good idea to not use default ports and not expose the port permanently but only when you need it. (Thats actually what I do, since I rarely need it anyway)


    But still this is just hardening against common noob attacks. To some level, you are still vulnerable to serious attacks. At least such attacks occur very rarely, if at all, unless you are a worthwhile target. To protect yourself against this while still exposing ports, you would have to go to a lot more effort. Eg. companies have much more complex security / firewall concepts. But this is hardly worth it at home. At home, the most efficient and effective solution for secure permanent access is to allow it only via VPN.


    Regarding your logs it is not the failed login attempts what looks strange for me, but the warning about IPs that have been added to your known hosts. I'm not a professional but could be that someone successfully did some nasty on your server. So I agree best solution may be to reinstall everything.

    ... So it is a completely OK state if one disk is missing,

    No it is not. You gettin that error due to misusage by not unmounting the device.


    Commenting out the lines in etc/monit/conf.d/openmediavault-filesystem.conf is not a good solution and so too, nothing else what silences the error messages would be a good solution but only a very bad hack of disabling an important system message that tries to tell you something.

    So what do you actually want to do?


    but when I want to create the disks in the File System

    You cannot create a disk in filesystem. That makes totally no sense at all. In the filesystems tab you can create a new filesystem on an empty / wiped disk. If you ve got a disk with an existing filesystem you just mount in order to use it or you wipe the disk and create a new one.

    Ist wirklich gar kein Angriff, nur als konstruktive Kritik gemeint. Was du dir jetzt genau zerschossen hast kann ich leider auch nicht genau sagen. Du könntest dich über Konsole direkt am System mal anmelden und die Logs durchstöbern. Evtl hilft dir der Befehl omv-firstaid um das Problem zu beheben. Bei sowas ist der Zeitansatz natürlich immer schwer zu beurteilen. Möglich dass eine Neuinstallation die bessere Wahl ist, wenn du keine große Lust auf Fehlersuche hast und die Konfiguration nicht zu komplex war. Ansonsten ggf. ein System-Backup wiederherstellen.

    Ich habs inzwischen "versaut".

    Hab unter Festplatte (glaube ich zumindest) "diese Platte" (schnell) löschen, für jede Platte, ausgewählt.

    Nun kann ich den OMV leider nicht mehr über den Browser oder SSH erreichen, nur noch direkt an der Maschine ist die Console verfügbar.

    Einen Tipp, wie ich (nach meiner voreiligen Aktion) weitermachen kann?

    Mir ist allerdings nicht klar, warum der Zugriff per Browser/sSH nicht mehr klappt).


    Syslog checken. Ggf omv-firstaid.


    Um ehrlich zu sein dafür dass du so rigoros mit deinem System umgehst und dir ggf Downtime nichts auszumachen scheint verstehe ich nicht, warum du unbedingt RAID 1 oder 5 benutzen willst. Immerhin kostet die Festplatte für die Parität bares Geld und bringt dir nichts weiter als im seltenen Fall eines Defekts Downtime zu verhindern. Gleichzeitig bastelst du aber mit dem Rohrstock an deinem System herum und zerschießt es dir komplett weil du einfach mal ohne Vorbereitung Dateisysteme entfernst. Das passt nicht zusammen. Da würde es mehr Sinn ergeben das Geld beim RAID zu sparen und dafür vorsichtiger am Server zu arbeiten. Mann kann auch in einer VM erstmal testen was man vorhat. Nicht böse gemeint.

    Das wird so nicht gehen, weil man nicht einfach Platten auswerfen kann die in Verwendung sind. Der einfachste Weg ist, als erstes die freigegebenen Ordner auf andere Orte zeigen zu lassen die nicht auf den Laufwerken liegen die du ummodeln willst (Also z. B. einfach auf dem Systemlaufwerk einen Ordner _tmp erzeugen und alle betroffenen freigegebenen Ordner darauf zeigen lassen). Erst wenn alle Verknüpfungen auf das Dateisystem aufgehoben sind steht unter Dateisysteme => Referenziert => nein. Dann kannst du das Dateisystem aushängen und anschließend das RAID auflösen.



    I also wonder if it's now possible to set custom mount options like noatime or compression can be set and persist after reboot without any error?

    Do you agree that linear raid should be removed from our beloved OMV?


    I think I am far away from being expert enough to give votdev any advice here. Perhaps the option is just there because he copied the list of upstream available options. So a better question may be why is it even possible on system level to create it? Yes I know, I said I think it's useless but to be fair, I did not even spend 5 mins to find out.. If you find sth, let me know.

    I am using stripe since it can bring higher performance. I think a linear RAID will not bring you performance or any other advantages. It is totally useless. Striping has pro of higher performance and con of higher risk that data is not available in case of a single drive failure. In my case the drives are the speed bottleneck and data availability in case of drive failure is not important, so Striping is definitely the right choice.


    We've had the discussion here many times, but I hate these statements "You can take striping if you don't care about your data.... If a drive fails you lose a lot". NO. I am never going to lose any data because of this, since data safety is not a RAID business but a backup business. If one drive fails, my data is not available anymore until I have fixed the hardware and recovered a backup. The only risk here is server downtime.

    With such a small amount of data and backups planned once a week only, I would not invest the extra money in a backup server but in additional backup drives to be stored at another place. As KM0201 already mentioned, what happens in case of a fire that destroys not only your NAS but too your primary backup drives? You could eg have a second (encrypted) backup drive at work or so. I think that is the most efficient way as long as you don't need sth like daily or hourly backups. Talking about daily or hourly, I would setup an offsite backup server.

    Yeah you should use docker. Just install OMVextras the same way as in OMV4 and there you can install docker. I would recommend using docker-compose as well, since here you can set all settings in a simple yml file and then just start with


    Code
    docker-compose up (-d)


    I recommend the images of linuxserver.io. They are highly maintained and configurable, well documented and I've never had problems with them. Their nginx image: https://hub.docker.com/r/linuxserver/nginx

    From the text I don't really get your point. But Port forwarding is done at your router settings. When you currently end up on secure OMV gui, you have probably forwarded port 443 to 443. Means what calls your public IP on Port 443 is forwarded to your omv-ip Port 443. If you want it to go to 8020 you must set 443 to 8020 in your router settings.


    If you want it to forward to nginx proxy you must obviously set 443 to what ever your nginx proxy is listening on and then setting the jellyfin proxy conf. I dunno if I'm just too stupid to get your point but for me it sounds like you should start with a more easy setup to get familiar with port forwarding and networking stuff.

    ZFS has a procedure for exporting a pool and then importing it into another system. BTRFS does not have a similar procedure?


    That is simply not needed. As soon as one disk is mounted, the Kernel recognizes that the filesystem is part of a BTRFS RAID connected to other drives and mounts the other drives as well. As far as I can say this is a problem of OMV6 only.


    There was another thread about it where votdev mentioned some problems but did not go into detail. I think it was sth about missing mount options when drives got mounted after reboot? So perhaps it is about saltstack? idk. However, still things about BTRFS support on OMV6 are quite unclear. Maybe votdev can clear things up.


    Edit: Thats the other thread: Is there btrfs support planned for omv6?