Yes I have max backup size set to 0.
Not disabling this would've likely prevented filling the filesystem.
Yes I have max backup size set to 0.
Not disabling this would've likely prevented filling the filesystem.
is the default configuration, meaning, if category.create isn't present in the configuration, it's still activated?
It is the default currently but trapexit is changing to a non-existing path policy as the default because existing path is support headache.
category.create is always present in the mount file - https://github.com/OpenMediaVa…stem-mergerfs_mount.j2#L7
Is this blocking epmfs from going on to copy on disk 2 since this directory isn't availabel there?
As far as I know it is.
So if i would create it manually, MergerFS actually would go on copying to drive 2?
Yes.
Isn't MergerFS supposed to go on copying on to the second disk?
It will if you aren't using an "existing path" policy. I recommend reading how the different policies work in the mergerfs docs - https://trapexit.github.io/mergerfs/
Or doesn't MergerFS work when accessing the mounting point on the cli?
As long as you are accessing the pool instead of one of the underlying filesystems in the pool, it works just fine from the CLI.
Solved this by using the # SKIP_BACKUP flag but not sure why this is needed when directories are marked RO and this has only recently started to happen?
I had a typo when making a change to fix another issue. It is fixed in the testing version - test new version of compose plugin that has a docker exec terminal
I really didn't know people were using :ro on so many things other than a single file. Do you have max backup size disabled (set to 0)?
It might be a while before it is released but it is in the testing package here
Change
Mailtransport is a key service. Notifications is a client using this key service. Allowing a client to turn off the key service would be like allowing an electricity consumer to turn of the power plant.
I would agree on a normal server but many things OMV does only has OMV in mind. It is very hard to accommodate services outside of OMV that may or may not be there. I would argue this would at most need to be an environment variable because very few people are going to not want any OMV notifications and want postfix running. I have never seen anyone mention this situation in 15+ years until now.
What's the prefered way (if any) to provide feedback on existing features and propose improvements.
Volker prefers github issues. But as I mentioned above, I prefer that things get hashed out here before Volker spend his time on a github issue.
I.e. I just found out that deactivating "Notifications" also stops postfix from running which I think isn't a good decision. I understand notfications need the underlying mail transport, but that does not automatically mean postfix should be turned off if notifications are not running. In my case, i have restic scripts running, sending out mails. So when Notifications are turned off, no mails from restic (also it's not transparent from the UI that notifcations also turn off mailtransport). So my proposal would be decouple postfix from notifications, either having postfix runing all the time or defining it as a service which can be started or stopped independently.
The problem is that some notifications will continue to be sent if postfix is running.
It would be strange if everything in OMV told you what it was doing on the backend in the web interface.
Do I need to edit the grub and add the GRUB_CMDLINE_LINUX="intel_iommu=on iommu=pt" or is that taken care of with the installation of the kvm plugin?
The plugin is not making any grub changes.
Maybe consult with chatgpt on the windows error.
t's precisely because Volker is the #1 contributo
I am well aware. He is basically the only contributor.
I am inclined to believe that he don't spend much time reading the discussions happening on the forum. Nevertheless, I value your opinion.
He reads it more than you think. I just prefer that ideas be filtered on the forum before possibly wasting his time. Glad my opinion counts. I spend as much time on OMV between the forum and omv-extras as Volker and have been for the decade and a half.
I'd like to think that Volker keeps in touch with the forum's moderator(s) to hear on key insights or interesting use cases.
Not really. I bring things up to him once in a while but not often.
A simple reaction remains easiest and most effective (effort wise) way for a user to provide feedback
That is fine. Keep doing it. I was just saying it probably won't do much.
I've never created libvirt hooks. So, not sure why those are needed. Can't really help with Windows.
do know that I had to install the virt-manager in the first place because I kept getting errors when trying to edit the XML using the KVM plugin. It just wouldn't let me do it without erroring out.
What was the error? Moving to virt-manager because of error doesn't mean the plugin has bug or is creating xml incorrectly.
using virt-manager really shouldn't be needed anymore. All of those links in my post were used to help create the xml that plugin adds. There are steps you have to do on your system that virt-manager doesn't do either.
is there an instruction how it is possible to mount several different usb-hdds (rorating backup) never at the same time but always in the same directory?
Nope because you will have to do it outside of OMV. OMV expects disks to always be mounted. The usbbackup plugin can use usb disks but it is always rsync'ing files and likely wouldn't work for you.
if you are using LVM for your targets, then you can take an lvm snapshot and backup from the snapshot. I would guess that is what qnap is doing. If you aren't using LVM, you would have to disconnect/unbind the initiator to get a good backup.
Do you want to backup at the block level?
Thanks for pointing that out. I think it should be fixed now. vim is working. I didn't have time to try the mysql/mariadb client.
Just a warning that that plugin hasn't been touched since the OMV 4.x days. So, it is very out of date and had issues even back then. I would probably move it from certbot to acme.sh if someone was going to work on it.
I strongly believe that these kind of clues/hints, although it may seem irrelevant, do provide a good idea about the people's interest and perhaps usability or usefulness of such features, and may help prioritize the efforts dev team put in their work
While that might be a good indicator for a large project with large dev team, I know I don't pay attention to them for omv-extras stuff. Not sure if Volker does for core OMV but since he is the only dev, I don't know that he needs help prioritizing. I think a good discussion with many people here on the forum is better before filing an issue.
so I installed it, but it gave me this message "failed to read from socket, connection reset by peer".
Then of course the "Kernel" option under system gives a "software error, press left mouse button to continue", and takes me back to the normal OMV page.
This used to be common when installing a plugin because it restarts OMV's engined. just hitting ctrl-shift-R "fixes" the issue.
The latest upload maintains a persistent connection making the terminal much better.
I assume this would mean that I'm unable to right?
You install it in the guest and the only requirement is that the VM is 64 bit
we have seen plenty of issues with the 6.12 backports kernel. I would consider using the proxmox 6.8 or 6.11 kernel that you can install with the kernel plugin.