Posts by mamema

    Hi,


    i've upgraded from omv 6 to omv 7 with some minor glichtes..... php fpm nginx issues and not running the fix script.

    But this was done at the end....

    So i had the impression everything is working.

    But on shared folders on SMB shares users which had the permissions on OMV 6 are now not able to create/delete files and folders.

    Of course nothing has changed on permission side. The user has the correct permission within the permission config of the SMB config in OMV 7 set as read/write.


    of course i can trigger an chmod on command line, but i rather would skip this, because then my backups got triggered and "another 10 TB" would be backed up, which would "too much".....

    especially as the only change was the upgrade step to omv7.


    what can i do without modifiying the until now unchanged file permissions on terminal level?

    Hi,


    i've updated from OMV 6 to 7 with


    omv-release-upgrade


    then had "all red" information in web gui, which i've fixed with:


    sudo omv-salt deploy run nginx phpfpm

    sudo systemctl restart nginx

    sudo systemctl restart php8.2-fpm


    but after login the KVM plugin throws errors (see topic in question) this occurs when selecting the VMs tab and the plugin "browses" for VMs.

    The list stays empty and the error is thrown.


    I don't see anythin g obvious in syslog.....

    In KVM log i see this:

    5/6/2024, 2:35:29 PM

    libvirt-php/connection libvirt_connect
    Connection to qemu:///system established, returning 0x55e56a569030

    5/6/2024, 2:35:29 PM

    libvirt-php/domain libvirt_list_domains
    Found 12 domains


    But web interface shows 500 error and stais empty.


    also started the web interface with a privagte web browser window. No difference.


    any hints?

    i was booting from an USB 2.0 device my openmedivault server since years without problems.

    Even though, it's said USB 2.0 is slow, it was okay

    Now, for housekeeping issues and the upcoming openmedivault upgrade i thought, let's buy a new USB 3.0 larger USB device. Doesn't cost much.....

    So, i bought a sandisk 128 GB USB 3.1 device

    Cloned by USB 2.0 boot stick from the backup taken from the backup plugin


    Booting from the USB 3.1 stick was slow as hell. WTF? Dashboard unresponsive and iowait was on 25! The USB 3.0 stick was slower than the USB 2.0 stick

    Does my older Server hardware not like this new stick?

    Surfed two days the internet and found a lot USB 3.0 slow on linux posts. also here in this forum....


    So i was starting the troubleshooting

    - is USB 3.0 enabled in bios? - yes

    - is USB 3.0 support loaded from the kernel ?- yes

    - does the situation changes on other USB ports? - no


    So i was booting from the USB 2.0 stick and added the USB 3.0 stick to compare the read/write speeds of the USB 2.0 & USB 3.0 device with hdparam.

    Uhh! The USB 3.0 stick was 15 times faster. So why is the booting so slow?


    Shortly after the hdparam tests, one of the devices went into read only mode. WTF part 2!


    But then it made click in my mind. As the devices where cloned with Etcher, they had the same UUID, so the OS had problems to decide, which device is the unique one, as they are not unique at all....

    Could this be the issue?


    YES!!


    Even though, when the USB 3.0 was activated alone in the server and there shouldn't be a uniqueness issue, after setting a random UUID with gparted, the USB 3.0 device

    booted fast and works as expected

    This UUID issue was never an issue with my old two USB 2.0 devices which where the same vendor and model and same specs, so i was able to switch between them back and forth.

    But now i had a new device with different specs...

    Somewhere deep buried in the linux system, the old UUID was linked to the old device, which made problems with the new one.


    So, i've learned something.

    as said in #12 kvm-plugin is installed and now the correct version. Why it was not the current one is also a mystery to me.

    As i was playing around with win 11, i was trying to install swtpm manually.

    So back then i recognized something is not working right, as swtpm wasn't available.

    I was not aware that the kvm-plugin takes care of that.

    So, eventually disabling backports, running apt-clean, enabling backports solved some errors, then after years...:-) of install time kvm-plugin is back up running with tpm support.

    Thanks for helping me solving this

    You are about a couple dozen release behind. That is your problem. 6.2.18 is current.

    see answer #9 i do not know why i'm behind as i'm pressing every update button....

    so now i have no kvm, i guess because of my now surfacing issue the update was never installed.


    i've tried several steps which you propose in similar situations. Eventually after enabling backports, the kvm plugin is installing again. But why i've needed backports is over my head....

    kvm plugin is on 6.1.22..oh this interesting. If i look into the gui, there version 6.2.18 ist mentioned as installed... uhhh....


    argh, silly me, i've tried to unistall kvm plugin and reinstall, now i have an error


    The following packages have unmet dependencies: openmediavault-kvm : Depends: libvirt-daemon-driver-lxc but it is not going to be installed Depends: swtpm but it is not installable Depends: swtpm-tools but it is not installable

    ...so!

    This was a hard one.... but solved... kinda.....


    Current Situation

    OMV6 has still problems to provide a fool proof way to create bridging network interfaces for Docker and/or KVM environments


    What "you" shouldn't have.....

    - more than one NIC

    - different NIC speed like 1 GB and 10 GB


    of course no one has this.... :)


    so at the beginning of my problems i used omv-firstaid

    This is always helpful if you have console access to the server, because everything network wise gets recreated. With a headless server you might have a running server without network access.


    then.... how do you create a WORKING bridge config. Look at the capital letters......


    Here, this forum is misleading to different (non)working solutions


    you can edit files in /etc/netplan

    YOU SHOULD KNOW WHAT YOU ARE DOING THERE.....BUT....IT'S STILL BETTER THAN THE OMV GUI....because after reboot you might have a working bridge config for Docker/VMs


    BUT.... perhaps not..... then you have to use the OTHER solution, which is burried in the forums and mentioned by the "forum wikipedia guy @


    ryecoaaron

    "

    His solution: create the network config WITHOUT bridge setting (normal networking) in the OMV6 gui, save settings. check ip, ping, internet etc .

    THEN DELETE the network config WITHIN the gui, BUT DO NOT apply settings, then create bridge config WITHIN the OMV gui and THEN SAVE CONFIG


    ..... then you have a working bridge config.......

    you're finished...... if you have only 1GB networks.....


    BUT, if you have a mixed environment with 10 GB for example, you might want to change the MTU size to 9000 (jumbo frames) on those links, sure, you can do this in the network area in the GUI. Do it there, but you might only get the MTU of 9000 on the br0/br1 interfaces and NOT on the physical network card itself.

    This would lead to network dropouts....

    So you have to redo the network stuff and set the MTU size from the beginning (eg. start over, see above) or you are using

    ip link set dev name_of_network_device mtu 9000


    on every boot.....


    so..... solved, but still "room for improvment"

    Hi,


    see topic in question. I had to rebuild before this. OMV6 was running fine back then....

    Have:

    - OMV6

    - VMs

    - docker

    - KVM plugin (cockpit removed)


    OMV host can ping: inet, vms, docker

    VM can ping host, internet, docker


    docker: container with host network can ping VMs, internet, docker, host

    docker bridge docker0 cannot ping internet


    i had (of course) problems to build br0,br1 (2 nics) with the omv gui, so i created manually within /etc/netplan

    br0, br1 are working


    docker0 is not.....


    as openmediavault is the "boss" on this server, i'm asking here, so it wouldn't break something if i'm fiddeling around at the wrong end......


    what is the best approach to get docker0 bridge working?

    have you ever thought to use netdata? Its also available as docker container. It shows everything in detail, also every docker usage

    i've migrated successfully to omv6 a few weeks ago, the following i never had with omv5


    since migration to omv6 several times a day, my netdata monitoring triggers a warning, as the ups goes to 90%, then gets back to 100%

    This cycle can happen several times a day.


    What i am unsure of and therefor ask the community is:


    - this was also the case in. omv5 .... it's just netdata which has improved?

    - this is new in omv6 nut plugin and normal battery management?

    - this is a bug?

    - my ups is dying?


    what is it?

    i got it solved. Was a combination of different things.

    First: the above config is working

    Second: i haven't changed haproxy configs for a long time, so i missed a "new" bug, which is, that the reload of a changed config no longer works. i have to reboot the whole firewall (runs with pfsense) to reapply the configs. Well, dunno if pfsense or haproxy is the culprit here

    Third: As photoprism isn't multiuser (yet) i had two docker containers running for two persons, even tough nothing in the logs, i had to seperate each port setting internally and externally, which means the standard ports in every config area had to be changed so they don't overlap.


    Well.... thanks for the help guys

    yes i changed this for posting (example.com) and yes http://.... is working. And yes haproxy is awqard. var... is for the 2FA config, which nisn't used in this example, as i first test inb internal network.


    The overall config "reversing....." is working, if i just change a backend to another one, i get to that other backend. Do i change it back to the photoprism backend, i'm landin on the omv login page.....

    okay you're right, but why it isn't working then? :)

    i'll explain the stuff for those not familiar with haproxy