Search results

Ping times are way too high for corosync. And do I read that correctly that there is an Internet connection involved? Do you use some kind of VPN between the sites? As can be seen with the round trip times (rtt min/avg/max/mdev = 15.667/25.435/33.892/3.466 ms) there is also much jitter on the...

What is the network latency between the nodes on the network used for corosync? is this network on a shared medium? Can there be congestion?

Do not use consumer devices with Ceph.

When even the kernel shows an IO error (in dmesg) it is time to replace the device.

After changing the hostname a reboot is recommended as other services may also still use the old hostname.

Please do not install packages outside of the Proxmox and Debian repositories if you not even know how to run the installer. Just my ¢2.

Das sind schon allein zu wenig Knoten für Ceph.

Wie sieht denn der Rest des Clusters.aus? Ich würde davon ausgehen, dass ein Pi zu wenig Leistung für ein OSD hat.

Was soll denn auf dem Pi laufen?

You can assign 2 or more votes to the Qdevice. This could help by providing more votes. E.g.: each PVE node has one vote and the Qdevice has 3. Total votes are 19, majority is 10. 9 PVE hosts can fail and there is still a majority of votes available. But if the Qdevice fails only 6 other PVE...

CVE-2026-32746 affects telnetd. Do you have telnetd installed or even enabled on any of your machines?

Warum sollte Multipathing den discard-Befehl nicht weitergeben? Bzw sollte das Discard ja die qcow2-Datei im OCFS2 kleiner machen. Auf der Ebene hat das mit Multipathing zum SAN ja noch gar nichts zu tun. Wenn das SAN-Storage selber die LUN thin-provisioned macht, kann ein discard aus der VM...

By default /etc/ceph/ceph.conf is a symlink to /etc/pve/ceph.conf which makes it the same on each cluster node. AFAIK it is easier to use ceph config set to set the values in the config db for each Proxmox node. ceph config set client.HOSTNAME crush_location az=az1

Würde ich so auch vermuten.

What is the network latency between the nodes, especially between the different AZs?

AFAIK NetApp has one of the fastest and most stable NFS server implementation in the industry. If you already have that I would definitely run some benchmarks on it.

And RedHat already has a virtualization offering for their enterprise customers. I doubt that anybody really entrenched into the "Enterprise" RHEL ecosystem is looking elsewhere.

We also use cloud-init and have no issues. The hostname was always set to the name of the VM. But we use an FQDN as VM name.

But it will only be one Proxmox node? And you need to mount the same filesystem in multiple VMs? Then ZFS with NFS/SMB shares on Proxmox.

Then I would suggest to use ZFS locally. Ceph is a distributed storage system that works best with five nodes or more.