Recent content by florian-n

Falk hat absolut recht - Netzwerkprobleme können zu einer OSD-degradierung (OSD = out) führen, die nur manuell aufgelöst werden kann. Siehe https://docs.ceph.com/en/latest/rados/configuration/mon-osd-interaction Ursachen hierfür kann sein, dass der CEPH-Traffic nicht über dedizierte...

Hallo @Winet.maier , auf den ersten Blick sehe ich hier zwei Probleme: 1. Die OSD muss neu gestartet werden (ist mir in der Praxis noch nie vorgekommen) und 2. der Cluster bleibt "stehen" sobald eine OSD vom Netz geht. Um letzteres wirklich beurteilen zu können, ist es wichtig mehr über den...

Hallo, das ist leider schwer zu sagen ohne weitere Informationen. - Welches Update wurde eingespielt? - Sind nur die OSDs des aktualisierten Nodes betrroffen? Falls ja deuten die Schreibprobleme auf eine fehlende Redundanz von Ceph Komponenten (https://docs.ceph.com/en/mimic/start/intro/) hin...

Eactly - i'd widen that answer: as long as you didn't aim for the explicit use of the RDMA-feature you likely don't need it, because it requires special switching capabilities (see f.e. https://community.fs.com/encyclopedia/remote-direct-memory-access-rdma.html). And unless you're trying to...

I was able to fix this by disabling the the RDMA-feature of both onboard adapters in the BIOS.

Thank you very much @LeMeow - this is a very nice appreciation! :D

It seems like its merged and comes with OpenZFS 2.3 (https://github.com/openzfs/zfs/pull/15022). ZFS 2.3 is not yet released, given the current release is 2.2.3 and Proxmox delivers 2.2.3 at the time of writing i'd expect this to land this autumn as soon as 2.3 is released.

This issue happened because the "node" property wasn't set properly.

Hi we're trying to deploy a linked clone via Ansible's "community.general.proxmox_kvm" module. The template vm (disk and cloudinit disk) is laid out on a cluster wide accessible RBD storage. During the creation step we receive a "500 Internal Server Error": Unable to clone vm fam-dev from...

It turned out i was missing a user change between the deployment of the template (using Packer) and the cloning of that VM template, which is being done using an API token. It turned out it didn't work (and couldn't see the VM template in the cluster) because the token missed the according...

I am experimenting a while with different configurations and setups trying to use a central VM template for the whole cluster. We are using a cluster with enabled Ceph integration. Currently the template VM is being installed on a RBD storage which is available to every node. Unfortunately the...

Thank you very much @TodorPetkov - this option was exactly what i was missing. I moved the cloud-init based attributes to the update-task. It was important to remove the state attribute from the update-task, i guess the "proxmox_kvm" module should have better eliminated the update attribute and...

I'm unsure if i'm on the right track. :(

Huh - i oversaw i had to manually add a cloud-init after cloning using the UI. I guess it might me be necessary to set the ci up afterwards. Thank you @TodorPetkov - sure probably you have more insights than me.. :) - name: Deploy continuous cloud machines community.general.proxmox_kvm...

Hi, currently we're trying to create template based clones using Ansible's community.general.proxmox_kvm. To make the clones configurable we stick with cloud-init. We are building the template using a hand rolled Packer configuration which deploys fine (locally as well as in our production...