dakralex's latest activity
-
For reference: https://bugzilla.proxmox.com/show_bug.cgi?id=6314 -
hi, at the moment there is no option to weigh CPU load balancing higher than memory usage balancing. feel free to file a feature request on bugzilla [0] [0] https://bugzilla.proxmox.com/
-
dakralex replied to the thread Questions about the dynamic CRS.This should be the same for VMs and CTs as the HA Manager handles affinity rules the same for both. In a quick test, the migration blocker with the same node affinity rule as above was shown to me for both VMs and CTs, so I couldn't move either...
-
It should not let you migrate neither the CT nor the VM. If it did, once you migrate a VM from pve-4 to pve-2, in the next ha cycle (~10 seconds) HA would say "this VM on pve-2 has another node with higher priority and the node is up, let's...
-
dakralex reacted to drwadu's post in the thread Question about Proxmox Cluster Resource Scheduling with
Like.
one thing to keep in mind about what causes imbalance and why migrations might not be effective is that non-HA-managed VMs contribute to imbalance too, but other than HA-managed resources they cannot be moved around to reduce imbalance your... -
dakralex replied to the thread Adding HA resources to an existing HA affinity rule and outdated manpages.This comment in BZ 7260 [0] does state how HA resources can be added by adding items to the existing HA resources list of the node affinity rule. However, is there a specific use case why this is done in the pre-start hook-script instead of the...
-
dakralex reacted to t.lamprecht's post in the thread Proxmox Virtual Environment now available for 64-bit ARM (arm64)! with Like.
We are excited to announce the first release of Proxmox Virtual Environment with official support for a second CPU architecture: 64-bit ARM (arm64/aarch64). Until now, Proxmox VE was available for x86-64 (amd64) only. Proxmox VE 9.2 for arm64...
-
dakralex replied to the thread PVE 9.1.11: HA CRM stuck in wait_for_quorum despite corosync quorum OK.Hi! This would be the expected behavior if there are no and have never been any active HA resource present (i.e. running) in the cluster. The CRM will only transition into an active state as soon as there are any active HA resource present, the...
-
dakralex replied to the thread Need Clarification on High Availability / CRS Behavior.Yes, to be exact rebalance-on-start will only be relevant if the HA resource goes through the request_start state, which can only happen if the HA resource was previously in the stopped state and e.g. a user requests the guest to be started...
-
dakralex replied to the thread Need Clarification on High Availability / CRS Behavior.Hi! In case of a fenced node, the HA resources on the fenced node are put in recovery state and the HA Manager will select a recovery node for each HA resource individually. This will always try to spread the HA resources on the remaining...
-
dakralex replied to the thread Questions about the dynamic CRS.Generally, these sound like good ideas! Feel free to create separate Bugzilla entries for the separate issues: automatically add new guests as HA resources by enabling some cluster-wide check (Though new guests can be automatically added...
-
dakralex replied to the thread Questions about the dynamic CRS.Hi Jonas! Good catch! There is a patch for this already on the mailing list [0], I will send a new revision soon so this gets resolved! This does only happen with the "migrate" shutdown policy and is documented in the documentation for the HA...
-
dakralex replied to the thread About CRS Dynamic-load Scheduler Status.If a system needs high availability, there must be enough redundant hardware to compensate such node failures across the cluster. How many cluster node failures should be compensated is a question of cluster size (i.e. what is possible with...
-
Hi! The section you're referring to states that the rebalance on start feature is in technology preview, not the CRS load balancer. The technology preview status for the CRS load balancer has been removed recently. But you are right, that the...
-
dakralex replied to the thread HA trouble with SDN.To clarify this here, this is the expected behavior for the given HA affinity rule set. Essentially, if the HA resource has failback enabled (which it has by default) and is part of a node affinity rule with node priority classes, the HA resource...
-
dakralex replied to the thread Cannot migrate VM, because HA resource vm is not allowed.Sorry for the late reply, missed this thread in the new posts section. This was already stated some of the other answers, but this is the expected behavior: HA resources are by default configured with "failback" enabled. Failback in that sense...
-
dakralex replied to the thread After PVE upgrade to 9.2.5 I cannot more migrate VM.Hi! If I understand your description correctly, this is expected behavior. As HA resources are "failback" by default and the HA resource is in a non-strict node affinity rule, the HA Manager does not allow the migration. This is because the HA...
-
dakralex reacted to j.theisen's post in the thread Powering down a Proxmox 9.2 cluster: how-to needed with Like.
Hi @engineer5 thank you for posting on the forum! So with the new HA disarm feature you can also chose to ignore the state of the VMs instead of freezing everything. This will still let you properly power down all the VMs. After the...
-
dakralex replied to the thread Questions about the dynamic CRS.Hi, sorry for the late reply! At least for two rooms this is already possible with non-strict node affinity rules as these will fallback to the other available cluster nodes if none of the specified nodes are available anymore. However, an HA...
-
dakralex reacted to aaron's post in the thread HA fencing with softdog vs IPMI watchdog — which one is actually used when both are configured? with Like.
I recommend to stick with the soft watchdog. It works well and you avoid any issues due to questionable quality of OOBM hardware and software ;)