I have decided to start fresh with Nethserver 8 and document my issues, feedback with the community with hope that I can find solution or bring them to light so they can be addressed in coming releases.
I am not sure if i want to open a new thread with each issue. Perhaps using this single thread to log the journey is better way to go?
I have created a fresh VM, installed Debian 12 then installed NS8 using the curl command as described in the wiki. After logging in the first time, i created the a cluster, then rebooted. I wanted to try the netdata as first app to install. Installation ended with error as following:
I tried to uninstall the app but again ended up with the following error. The app remained uninstalled. Even after several attempt the app remained uninstalled.
I did take a snapshot of the vanilla debian so i can always revert back. But the question is why the fresh nethserver 8 with nothing confgured at all have issue installing a simple app such as netdata? Any idea?
Thanks for starting over and explaining the issues.
I think that the relevant error information is hidden behind the âShow moreâŠâ links (bottom right of the error textbox) in the first screenshot.
It should be possible to get it from the system logs page.
The best way to keep it clean is one support thread for one issue, also because itâs better for finding solutions and reading in general.
Just an idea: Maybe thereâs a conflicting cloud-init config?
Here is a screenshot of error when trying to uninstall the netdata app. I have clicked on Show More , so both details are showing complete info. It does look like the issue is due to incomplete Install which was the original issue. How can I remove faulty install?
I am not adding much to the issue at hand, but it strikes me that many issues (not just this one, but many more) seem to be Debian related. I simply thought I mention this, thatâs all. No flaming intended here at all.
Ok, dumb section hereâŠ
On NS7 NetData is a wonderful tool to understand how data flows in and out from the OS for every service.
Into a container⊠which âojectâ's traffic is monitoring?
Since NS8 is all container based, technically OS should not matter. I think that was one of the reason to go with container.
An update on the netdata broken install, I am surprised myself. So if anyone got any explanation for this I would greatly appreciate it.
I could not uninstall the netdata no matter what. Installing new instances also was unsuccessful. I always use ext4 file system for OSs. Since I am trying many things to see what could be causing the issue, I decided to try btrfs instead of ext4. NS8 GUI did seem little more responsive to me, and then to my surprise, netdata installed on first try. Then I installed SAMBA and that install worked on first try as well. I was going to open a new thread for samba issue and post full error log because that also had unsuccessful install issue.
I should mention here, beside the btrfs in the VM, everything else is identical as before. VM stored on same storage, same configuration, same steps to install NS8 and netdata as first app install.
I am going to try to install mail, webtop and nextcloud next and see if they work. As I mentioned I am finding the GUI more responsive than before, enough to be useable.
I know btrfs is the faster file system, but why that should make such difference for podman based container? Anything in the neth design that makes btrfs suitable? I use debian for everrything and dont want to try anything else. So all my tests are on Debian 12.
At the moment, NethServer 8 seems to be the only candidate in the Debian universe having basic problems even installing on bare metal? Proxmox, OpenmediaVault, Ubuntu and others are all based of Debian - and all can be installed without any issues.
Yet NethServer 8, at the moment is VERY flakey on Debian, a little less on Rocky Linux.
Seems Rocky gets more support, dev resourcesâŠ
I really wonder why Nethesis insisted on using and thus also supporting âNobodyâ LinuxâŠ
Gregory Kurtzer isnât really known for creating anything open source, just as acting a parrot for codeâŠ
I am not really happy with the situation at the moment.
I will be testing another Debian based VM tonight (Different environment / Client), will see how it goesâŠ
I run for example an NS8 Debian 12 cluster using 3 nodes without issues for some time now with many installed apps including community ones. 1 on Proxmox at home (local NVME) , 1 Strato VPS, 1 Contabo VPS.
Debian for example uses an older podman version so there may be issues but so far everything is working here.
Iâd really like to find out why these issues occurâŠ
I can fully understand that a not working system doesnât make one happy.
Did you already try an installation on a local disk on the node to check if itâs maybe a NAS issue?
Just for the heck of it, I tested again with Debian+ext4, and all the issue of bad app install returned.
Looks like if NS8 used Debian, then btrfs should be used. For the most part this seems to be the solution for Debian users. Some issue still occurred such as WebTop DB issue i mentioned in the other post, but it is usable.
It is a valid point really. But the containerization should mean it works on all(most) flavours. For somebody like me who are reluctant to use any other OS except Debian, it does provide option to use NS8. After CentOS lost itâs charm, I am not all for going with rolling centos flavours.
I also follow the same ideals⊠Nothing RHEL based for me anymore - I just donât trust them anymore, they are on the same level as anyone singing the âbluesâ (Big BlueâŠ).
I had the same issues with XFS as file system I trust more than ext4, although I also tried ext4, both gave the same results.
Iâve never really used Debian with BTRfs, although I use BTRfs on a lot of systems, amoung them Synology NAS and a few Linux variants (Rock Store)âŠ
My fav is still ZFS, but only if the system fully supports it. I wonât attempt to run a system on ZFS if it doesnât support ZFS out of the box.
One thing that still makes me stop from using BTRfs is the fact that Proxmox Backup Server canât really provide file / folder level restore with BTRfs, as itâs not supported there. Ext4 and XFS are well supported.
And all machines run as VMs in Proxmox for me & my clients.
The only exception to that are 2 Novell Netware servers, they are running on VMWare ESXi (supported!), which itself runs in Proxmox (Yes, that works rock solid!). This is only for historical / legal archival reasons!
You were careful enough to use the âshouldâ formâŠ
I have never used netdata on NS7. So I am not the most experience one to answer. But from what I can tell, it looks like itâs monitoring the VM/machine itself.
There is a dedicated section in netdata named Container and VM and that seem to show data related to container environment. It would be good to know if netdata can show stats for individual containers to see which app using how much resources. Perhaps the options are there and I need to spend more time with netdata.
This is probably digresses the original topic of this thread, but that right there is an absolute crucial piece of information. I also run 100% Proxmox environment with heavy proxmox backup server. I am running a backup right now to see it for myself.
I dont want to say this will be a deal breaker if that is the case. Because I am not sure how much file/folder I do have access now inside the container environment. Say I deploy nextcloud on NS8, how much visibility do I have inside the container with PBS or without.
You wernât aware of the caveat with BtrFS & PBS?
I became aware of that when virtualizing MacOS with PBS as backupâŠ
Windows NTFS is not a problem for PBS.
But as it is mission critical for me, I donât trust OS backups too much for disaster recovery, or for file / folder fast restore, wheras PBS does cover that.
Note: I usually do 4 different Backups: Proxmox PBS, Proxmox NAS, OS Backups and my own Rsync script, grown after more than 20 yearsâŠ
PBS File / Folder restore works AFAIK for Containers too, as for LXC containers on Proxmox. PBS & Proxmox need to be able to handle the file system, that seems itâŠ
(HFS+ or BTRfs could be installed on bothâŠ)
My strategy:
Better doing a backup too many, than one too little!
My 2 cents
Andy
Itâs also possible that Proxmox adds these additional file systems, these guys have been moving fast in the past years!
@Andy_Wismer I just tested the Proxmox backup server for NS8 on Debian+btrfs. I can browse and restore file/folder without issue. I have two disk in the NS8. 2nd disk mounted as /home and BTRFS. I can see the folder structure and restore a single file from the backup.
Could you please confirm if I misunderstood you regarding BTRFS and single file restore on PBS? My PBS version is 2.4-6.
