Search results

Hi! Seems like there was something in the way of writing it to disk, you can try to gather the log on a separate console, e.g. netconsole [0], so that the information persists if there block io is unavailable. [0] https://www.kernel.org/doc/Documentation/networking/netconsole.txt

It would be helpful for the kernel stacktrace to be a bit longer, e.g. upload a whole boot and crash log (journalctl can also export a log from a whole boot). I can only see that ZFS seems involved, maybe some ZFS I/O worker panicked? Which CPU is installed on that system?

Either way, I'd first check if your hardware is alright, e.g. all cables and the RAM modules are connected correctly and aren't loose, there BIOS configuration is reset (no overlocking, etc.), and so forth. If it doesn't fix the problem, then a full boot log would be beneficial in finding the...

Hi! journalctl with PAGER set will usually open in that pager (typically it's less) and therefore you need to page down for journalctl --since="1 hour ago". For a better view of the log from the end of the last boot, you can run journalctl -b -1 -e. What is the output there?

From the provided log in log-pve540-1.zip it also seems like there's quite a few connection issues, e.g. to the InfluxDB right at the start. I assume that proxmoxW540 is one of the nodes that has the problems starting up? Is the network stable for that/these nodes?

Are these all nodes in the cluster? What does pvecm status output? Keep in mind that it is not recommended to have a even amount of nodes in a cluster, see e.g. [0]. This seems like a clustering issue first of all. Are the HA resources with node "---" running somewhere unnoticed? Or were these...

Hi! x86-64-v4 is not a CPU model configuration that QEMU directly provides, so it only works by specifying it for the qemu-server's cpu config, but not in args. What do you mean by "some flags aren't found"? It could be that these features aren't supported by the host CPU itself. If you want...

Hi! In general, the guest's operating system will usually pin each queue to a vCPU core. Therefore, there's not really a performance benefit to use more queues in total than vCPUs as then a vCPU must handle multiple queues then. Multiqueues are best experimented with by measuring the actual...

Hi! Are there any errors shown for Ceph or when listing images manually for that rbd storage?

Hi! It's best to show your interest in a specific Bugzilla entry so that we can track the interest in certain features. There's for example [0] for the general interest of adding any SMBIOS type information and/or [1] for automatically generating these serials. There's already support to...

Hi! Is there any syslog entry on the host or in the Debian VM when rebooting from inside the VM?

Hi! With GET /cluster/ha/rules and GET /cluster/ha/rules/{ruleid} you receive a digest value (of the whole config). If you provide this digest as an argument to PUT /cluster/ha/rules/{ruleid}, then the API handler will refuse to make changes to the HA rules if the digest (i.e. the rules...

The former one is just an alias for the latter one [0] ;) [0] https://manpages.debian.org/trixie/apt/apt.8.en.html

This fix is already done upstream [0] and for our patched ifupdown2 packages as well before Proxmox VE 9 was released, so this can only happen if the ifupdown2 package is somehow held back (e.g. using apt upgrade instead of apt dist-upgrade, which have different semantics how to upgrade). Make...

Hi! Um welches Netzwerkinterface handelt es sich? Wenn der Intel e1000e-Treiber verwendet wird, hilft womöglich die Information in [0]? Wenn die Fehlermeldungen direkt auf den Terminal tty0 gehen, könntest du einen anderen Terminal verwenden. Normalerweise wird stdout nicht als stdin...

Das wird durch den Fencing-Mechanismus sichergestellt: Standardmäßig verwendet der HA Stack das softdog Kernel Modul, welches unter Normalbedingungen in regelmäßigen Abständen diesen Watchdog resettet. Das kann aber nur passieren, wenn das LRM der Node auch das Node-eigene LRM-Lock besitzt...

Hi! Wenn "im Netzwerk isoliere" bedeutet, dass die Node keine Verbindung mehr zu den anderen Nodes in dem Cluster hat, dann würde das bedeuten, dass dadurch die ganze Node gefenced wird auf der die HA-Container laufen. Das bedeutet dann wiederum, dass auch Nicht-HA Gäste durch das Fencing...

Hi! Proxmox distributes its own patched ifupdown2 packages, so it's best to check if these issues are because of changes done there first. Either way, it should be checked out properly by creating a Bugzilla entry for this. Also the ifupdown2 version is from May 2023, while the most current...

From the apt history it looks like there was an in-between upgrade to many packages from Proxmox VE 9: Start-Date: 2025-10-05 16:21:44 Commandline: apt upgrade Install: [...] proxmox-kernel-6.14:amd64 (6.14.11-3, automatic), libjs-bootstrap5:amd64 (5.2.3+dfsg-8, automatic)...

Hi! I haven't tried to reproduce it yet: When was the update performed? Has the watchdog been inactive before already (some recent entries in journalctl -u watchdog-mux)?