Search results

Yes, apt dist-upgrade/apt full-upgrade must be used instead of apt upgrade as described in [0], because among other things the latter will make packages stay on a lower versioned package instead of resolving the package dependencies correctly. On which versions were you on Proxmox VE 7, 8, and...

Hi! What network does corosync run on? Does it have a dedicated network interface or is it shared with some other network resources? These are quite high numbers, the network should have a very low latency. [0] This seems like that for some reason the HA was interrupted while reading/writing...

Hi! Could you post some syslog on these nodes? Is HA setup on these nodes?

Does the performance improve if you upgrade to a newer version of pve-manager? There were some performance-related improvements in pve-manager >= 9.0.8.

Hi! Which exact version of pve-manager have you upgraded to (e.g. pveversion)? Could you post a syslog from one of the nodes you are experiencing the problems on?

Hi! What is the config of such a VM that fails to migrate (output of qm config <vmid>)? What does the tasklog show if the VM is migrated through HA? What does it show if the VM is migrated manually (does it work that way?)?

Hi! Which web guide are you referring to? HA groups have been deprecated in favor of the (semantically) equivalent HA node rules in Proxmox VE 9. The reason was primarily to aggregate those relationships in a single view, allow further extensions and make the interactions between node affinity...

Hi! Have you upgraded the Proxmox VE host from version 8 to 9 or is it a new install? Either way, HA groups have been deprecated in favor of HA node affinity rules and these are feature-compatible, even though they are presented slightly different. This includes that the "nofailback" flag is...

Hi! Thanks for the input! The linked feature request is still quite general in its description, but it is focused on the dynamic load balancing of HA resources first, i.e., the storage must be shared across the cluster nodes (or HA node affinity rules must be setup to ensure that a HA resource...

Hi! Which version of pve-manager and pve-ha-manager are you running on the nodes in the cluster? When the pve-ha-manager package is upgraded (which is dependent on upgrades to pve-manager), the LRM is put in restart mode and the local HA resources are then put in freeze state by the HA...

Hardware has wear too as any other component and sometimes that is even exacerbated by implementation faults, e.g. the 13th and 14th Intel 700 and 900 series had problems with overvoltage, which could render cores permanently degraded or even fail them entirely. Another cause could be that a...

Which CPU tests have you made? A good test suite that usually shows some signs of hardware trouble is stress-ng as these kinds of errors are usually caused when there's quite a load on the CPU. Otherwise random segfaults of widespread executables (such as perl, dnsmaq, python, ...) on a stable...

Hi! I couldn't fully reproduce the issue yet, but from what I can see there is a little difference in migration times for VM 910 (25 secs) and VM 911 (45 secs), which might cause the HA Manager to detect a separation between those guests inbetween and request that VM 910 should be put pack...

Hi! Thanks for the report! I'll look into it, this should be rather straightforward to fix. The package should be able to run independently on a machine without any Proxmox VE packages installed, but it seems that there are still some dependencies left. If you want you can create a Bugzilla...

Hi! Thanks for the feedback! Correct, HA resource affinity rules are currently strict only as one of the initial main use cases was to provide strict separation in case of a resource limitation, e.g. a specific PCIe device is only available once on each node and so guests must not be on the...

Thanks for the report! The other bug is already fixed in a newer version and for this bug there's now a proposed fix [0]. [0] https://lore.proxmox.com/pve-devel/20250905101648.79655-1-d.kral@proxmox.com/

Hi! Have you moved all guests from the host and powered off the host before issuing the pvecm delnode $nodename command [0]? What is the output of pvecm status and an output/excerpt of journalctl would be helpful to diagnose the issue further. [0]...

Hi! For larger clusters, we usually recommend a proper Ceph setup, but for smaller setups a NFS can do OK as a shared storage. For the rest of the post, it is hard to tell without any configurations and when the error exactly happens. How is the storage setup in PVE? Is it available on all...

Hi! AFAICT this is something that could be implemented in a rather straight-forward fashion, so if you want the feature, you can create a feature request at our Bugzilla [0]. If I didn't oversee it, there was no entry with that feature request yet, but please make sure to double check so...

Hi! Ein shared storage ist die beste Variante um HA zu garantieren, aber es ist auch möglich mit den Replication Jobs HA zu konfigurieren, wenn auf den Nodes jeweils das ZFS Pool auch den gleichen Namen hat und die Gäste so häufig wie möglich repliziert werden, um Datenverlust und/oder...