Posts by johanvandyck

    Hi


    I don't know if it can help: beginning of OMV5, netplan is used. The gui is creating files in /etc/netplan. These files are loaded into

    systemd-networkd.


    omv-firstaid is a textual manner to create new files in /etc/netplan. I don't know how docker is handling the network creating, but no doubt it creates an own network.


    Have look in netplan and try these commands to check and revert a netplan, generate the resulting set of the network and apply the new config. The last command reloads the netplan into the network.

    Code
    netplan --debug try
    netplan --debug generate
    netplan apply
    systemctl restart networking

    Finally I can tell the neplan files are executed from a till z. The last config overwrites the first one. You may safely add a file to the end: e.g. 60.yaml.

    Hi


    with 10gig nics, several items are very important:

    - jumboframes must be set to as high as possible in your entire network: nic and switch. size e.g.: 9k

    . This item alone, can double to tripple your speed.

    - offloading could help.

    - Can you describe how you test your speed: is the nic the problem, or the HDD/OS combination after it?

    Hi


    May I suggest to give some basic info? your system, your OS, your kernel, OMV version,...


    error: what is dmesg telling you when you insert your nic-to-usb? Can you give me some output


    If no firmware driver is available, you need to check if the kernel module is available and loaded.

    Code
    sudo modprobe -r r8169

    You could try: or something simular

    Code
    sudo insmod /usr/lib/modules/5.8.11-1-MANJARO/kernel/drivers/net/ethernet/realtek/r8169.ko.xz

    and like always: google is your friend.


    cheers


    Johan.

    Hi


    First 2 points accomplished:

    - I like to block regular access and only keep the tagged vlan3 traffic.

    - I need to bond 2 nic on this vlan3. this can't be done via gui. LACP protocol used.


    1) first point: my bonding port bond0 has no explicitly assigned ip address. See config below. In theorie a hacker could gain access on my terramaster, reconfigure my network and assign himself an regular address. It is not blocked on switch level. (If I do so, the vlan 3 subnet won't work anymore)


    2) second point: Define on switch level a LAG with 802.3ad (also called LACD) support.

    3) for backup purposes, I keep a simple config created via the omv5 gui. This will be overwritten afterwards.

    I created a 60-myconfig.yaml file. This file is executed after the other one. Pay attention, you may no longer apply network config via the gui. It will may clear /etc/netplan/*


    execute and apply network:

    netplan --debug try

    netplan --debug generate

    netplan --debug apply

    systemctl restart networking


    The last point is remaining. Keep you posted.


    And my original question: still stands. omv doesn't like vlan tagging on my switch. I found a workaround by using bond0.3 subnetting without giving the bon0 interfaces an IP in the main subnet.

    Hi


    I did some more research:

    - complete new install starting from my subnet (192.168.30.x): on vmware level everythings works fine. After that I need to convert to an img file, write it to my SD-card and put it on my omv nas.

    - I connected a laptop to my omv network cable to check if my switch is configured well.

    - my switch (zyxel 1900, L2 switch) is setup to tag all packages with vlan ID 3.

    --> conclusion: I have the same issues: on ssh shell: dns is blocking; omv is not reachable from outsite; I can get dhcp addresses.


    lesson: I learned omv is not blocking some IP's / IP's are not hardcoded insite. However, vlan tagging is not working as expected on omv.


    Second experiment:

    - create nic enp2s0 without vlan: 192.168.1.12 using omv-firstaid

    - create vlan port on enp2so.3 (192.168.30.17)

    - config switch to be on default (no vlan) network, be a trunk and let pass tagged ID3 traffic.



    conclusion: both interfaces are working fine. Nice. It seems tagging traffic down the road on the switch is not a good idea with omv. However it works fine for other devices.


    to be continued:

    - I like to block regular access and only keep the tagged vlan3 traffic.

    - I need to bond 2 nic on this vlan3. this can't be done via gui. LACP protocol used.

    - I have a third 10Gbit nic which should directly connect, without switch, to one of my PC's.


    Johan.

    Hi


    Maybe you cans describe exactly what you like to configure. Or what problem you encounter. The OVM5 gui let you create bonding out of the box. And via ssh interface you can extend the netplan config file with your own config.


    Can that help?

    Hi


    Problem: after changing subnet of omv5, the external connection to my nas and dns is no longer working


    - I installed the latest omv5 on SD card (terramaster nas F4-422: intell cpu, 12Gb ram and 3 nic's insite) by creating an vmware image, convert the image, update grub, update os, (...) I used DHCP address with fixed lease of 192.168.1.12/24. Gateway and DNS given by dhcp: 192.168.1.1. Everything is working fine.

    - I switched my server subnet 192.168.30.x/24 (by gui or omv-firstaid).

    - Reboot.


    Problem:

    - DNS no longer working: on ssh prompt nlookup http://www.google.be -> no server found.

    - external connection is no longer possible (e.g. ping)


    However:

    - after reboot: I got my dhcp addresses 192.168.30.12

    - after reboot the ssh message is telling me my gui is available on 192.168.30.12:80

    - on the ssh prompt I see my gui is listening on *:80. with telnet I can check this: ok. site reachable


    What I tried:

    network:

    - omv-firstaid: nic fix addresses. somethimes it gave me errors back, sometimes not.

    - omv-firstaid: reconfigure gui port

    --> I checked /etc/netplan/* and applied netplan. It is not here where it is going wrong.

    - restart nginx, reconfigure nginx: "omv-salt stage run deploy"

    - I make sure apache is off.

    - iptables => alle chains accept. No firewall configured.

    - routing: my routes for the new subnet looks fine.

    DNS:

    - manual overwrite dns in /etc/resolv.conf by adding my dns server


    I slowly realised it is propably not the omv install itself, but somewhere in my OS level, my subnet is hardcoded. When I go back to the old subnet, everything is working fine.



    Can someone point me out where else I can dig?;(


    regards


    Johan.