Search results

Thats true, spinners are only usable with WAL/DB outsourced to NVMe (which means total destruction of all OSDs on NVMe fail). It's important to use high end SSDs with ceph, there are lots of tests in the ceph-users ML archives.

Your storage went away during runtime. Are you sure the device is in good condition?

Bist du sicher, dass du eine freie IP verwendet hast? Connection refused bedeutet, dass der Server erreichbar, der Port aber zu ist. Gleichzeitig berichtest du, dass dein Ping nicht funktioniert - das klingt schon sehr merkwürdig (oder du hast deine Firewall vergessen abzuschalten).

There is a difference between used space and cache space. ZFS does not need 50% but it will use as much free space as available (empty RAM is wasted RAM). 50% would mean 16gb RAM für ARC which is good but does not need to be this large, 4GB in total is perfectly acceptable for a machine that...

Can you check if your watchdog device has been enabled with "nowayout"? That would explain why the system has been forcefully reset, even if the daemon itself disconnects gracefully. The message itself looks like a classical watchdog timeout. I don't know the specific details of the PVE +...

Your log starts AFTER the cause, you need to attach some minutes of log before your snippet. Actually I don't think it was remotely triggered (which is possible). Is the server located at a trusted location or is it possible someone pressed CTRL + ALT + DEL, pressed the front button, issued...

With other words, KVM itself does not implement L1TF fixes for example? I've already set these params on both host and VM, I should be fine then.

Ok, das wundert mich dann doch sehr. So spontan hätte ich da auch keine weitere Idee. Mal dmesg und journal beobachtet? Für ZFS gibts übrigens ein Stat-Tool, dass die Performancecounter live anzeigt, vielleicht hilft das. Voller RAM heißt i.d.R. Pagecache, steht in top ein hoher Wert bei Dirty?

Hi! On a trusted system without multiple tenants it is safe to disable all CPU flaw mitigations. I already set and run these kernel params on the host: https://make-linux-fast-again.com/ Which CPU flags need to be set / unset under Hardware -> CPU for the VM? Based on the wording, I would for...

Thank you for the feedback. Did you enable the firewall option for your VM network interface (right there were you can set the hwaddr)? I was only able to reproduce this when the firewall option is not enabled.

Hast du bei dem Versuch denn die Aktivierung zwischendurch neu gemacht? Wenn die ein mal raus ist, bleibt sie das auch. Ich habe hier keine Probleme bei der Migration, allerdings verwende ich Libvirtd und nicht Proxmox.

Das ist der Auslöser, die Extrem-Variante. :-) Intel vs. Intel reicht übrigens auch nicht, das passiert sogar zwischen Xeon's verwandter Serien, wo sich nur der Takt unterscheidet. Man kann das Ganze umgehen, wenn man kvm64 verwendet, dann ist die CPU komptibel mit allem. Eigentlich sollte das...

You can check if you have this file on your system: /etc/apt/apt.conf.d/20auto-upgrades If yes, check it's content against this article: https://wiki.debian.org/UnattendedUpgrades You should definitely enable persistent logs: https://gist.github.com/JPvRiel/b7c185833da32631fa6ce65b40836887...

Did you enable unattended-upgrades on the host? It's possible that an upgrade caused this but not the update of repository feeds. Just browse "journalctl -r" back to 17:13 and you will find your cause.

Hast du mal probiert für den Pool "sync=disabled" zu setzen (bei der VM reicht Writeback)? Die EVOs, also sogar die Billig-Variante der Consumer-SSDs, sind wirklich die schlechtesten SSDs für solch einen Zweck. Tritt das mit den besseren PM983 auch auf?

Anderes CPU-Modell und in PVE unter Modell "host" ausgewählt? In diesem Fall kann das schon mal passieren.

Hi! Is it possible to merge an existing ceph cluster into a proxmox-cluster, where ceph is not configured? I am operating a rather large, standalone cluster with mixed clients (KVM using CentOS, Fedora and Debian), that I would like to extend by using the hyper-converged approach (creating new...

Looks like your node is under heavy load, maybe something that screwed up. Do you have "atop" on that machine? "atop 1" might lead you to the process in that case.

Hi! Fresh install on Supermicro X11DPi-N(T) using Proxmox v6.1. This message is looping through the log and saturates one core: [ 911.977646] i40e 0000:60:00.1: Error I40E_AQ_RC_ENOSPC, forcing overflow promiscuous on PF [ 955.642636] i40e 0000:60:00.1: Error I40E_AQ_RC_ENOSPC adding RX...

Are you sure your connection lasts more than an hour? Using the latest kernel causes a lot of issues with OVS an Proxmox for me (where CentOS and Debian stock work perfectly fine).