Search results

NAT automatically does MAC translation, how else should that work? You probably have to spoof vmbr0's MAC address for the one that's registered with your provider.

Ich verstehe den Sinn dahinter trotzdem nicht so ganz. Die Restore-Geschwindigkeit hängt noch an so viel mehr Faktoren, welche wertvolle Information erhoffst du dir von der Größe des Backups? Es dauert so lang, wie es dauert. Außerdem kannst du inzwischen live restoren, damit wird die...

You don't need to define an address at all, if you want all of the host's traffic go through the OPNSense VM, only a bridge without any further settings. The address for OPNSense's WAN is then defined inside the VM. There you have to choose "static address", like you would do with any lan interface.

Can they ping on the Corosync address?

You don't need an external stonith procedure. PVE works with a software watchdog and has a usual downtime in case of a node failure of 2 to 5 minutes. Additionally, if the node just doesn't react anymore, power consumption could be unchanged and you would never know that the node failed.

Ah, now I get the problem. Can you double click on that disk and uncheck the box „Backup“?

Well, with three nodes you're on the minimum. If one fails there is no other node to take the PGs and Ceph has to put PGs on hosts twice, which it initially tries to avoid. But if you keep that in mind and don't fill the OSDs for more than 66%, it will work.

Go to the hardware tab of the VM and click on Add->Harddrive at least three times. :)

And the files inside files.local indeed get backuped? Or is there only the folder after the restore?

Always run apt full-upgrade.

There are lots of guys with more experience in the hardware section. My guess would have been excessive IO but that doesn't seem to be the case.

Hm okay, three operations in parallel should indeed not be an issue.

Are there other VM images running on the disk when you do that? If so, how many?

This has always been a matter of seconds here ... What disk do you use?

It's still a lot slower and still has to be mapped in RAM, so more RAM is still the optimal solution.

Without deeper knowledge I would say that vmbr42 is acting like a switch and thus only forwards the traffic to the respective guest which is the valid recipient, just like a hardware switch would do. But I don't know an answer to your question.

That's because the firewall does its job: dropping packets. If you want pings, you have to allow ICMP packets, incoming and outgoing depending on the policy that you defined for host and guest.

You can't, pve-firewall is stateless.

As the error says: the boot medium seems to be corrupt. Check the checksum of your iso file and download and/or burn it again.

Good to know, thanks for sharing the solution.