Replication Fail

[croaky]

I have created 2 Jobs 1 for a qemu vm and 1 for a container
both failed and worked first after i moved the config files. when i try to login via ssh from one node to another everything works fine and i can manage the other server.

Any Idea what i did wrong ? All Nodes are in the Same IP-Range


Logs:
Proxmox
Virtual Environment 6.4-5
Container 200 (ct_ubuntu) on node 'pve1'
Logs
()
2021-06-01 14:11:01 200-0: start replication job
2021-06-01 14:11:01 200-0: guest => CT 200, running => 0
2021-06-01 14:11:01 200-0: volumes => local-zfs:subvol-200-disk-0
2021-06-01 14:11:01 200-0: (remote_prepare_local_job) ssh: connect to host 192.168.x.x port 22: No route to host
2021-06-01 14:11:01 200-0: end replication job with error: command '/usr/bin/ssh -e none -o 'BatchMode=yes' -o 'HostKeyAlias=pve2' root@192.168.x.x -- pvesr prepare-local-job 200-0 local-zfs:subvol-200-disk-0 --last_sync 0' failed: exit code 255

The Virtualization is available when i move the config File via mv /etc/pve/nodes..........


Votequorum information
----------------------
Expected votes: 3
Highest expected: 3
Total votes: 3
Quorum: 2
Flags: Quorate

 

[fiona]

Hi,
does the error still happen when you try again? Dumb question: is the shown IP correct? What is the output of /usr/bin/ssh -e none -o 'BatchMode=yes' -o 'HostKeyAlias=pve2' root@192.168.200.221 -- echo "works"?

 

[croaky]

Hi,
does the error still happen when you try again? Dumb question: is the shown IP correct? What is the output of /usr/bin/ssh -e none -o 'BatchMode=yes' -o 'HostKeyAlias=pve2' root@192.168.200.221 -- echo "works"?

Hi Fabian_E thank your for your answer.

I have tried several times from one machine to the other and backwards. I have also tried to replicate from the second node back to the first as the job changes automatically. For some reasons it does not work.


On Node 1:
Error Connection error 595: No route to host

After i move the Configs and restart the first Server the Summary is still green

 

[fiona]

Hi Fabian_E thank your for your answer.

I have tried several times from one machine to the other and backwards. I have also tried to replicate from the second node back to the first as the job changes automatically. For some reasons it does not work.

View attachment 26609

The replication jobs seem to be ok. What do you expect the replication jobs to do? They only replicate the guest's images to another node. If you want the guests themselves to be automatically migrated/recovered upon node failure, you need to configure HA.

 

[croaky]

how embarrassing i misunderstood the concept. i actually wanted a HA for local use so that if one node fails the VM resurfaces on the other one automatically.
A HA would probably not make sense in this case with replication.

 

[fiona]

You can use HA together with ZFS replication, but be aware that you will loose the data since the last sync in case of a node failure. That's why using shared storage is still the recommended way.

 

[croaky]

You can use HA together with ZFS replication, but be aware that you will loose the data since the last sync in case of a node failure. That's why using shared storage is still the recommended way.

Ok then a live migration in that sense is not possible either.

• ZFS Replication = When u don't have a shared stroage or to keep the migration time short. And for Disaster Recovery
• Sharad Storage to keep all VM Data into a Backup for Restore and else
• HA is practical but only if you can guarantee the redundancy of the hardware. What can cause more costs

 

[fiona]

Ok then a live migration in that sense is not possible either.

• ZFS Replication = When u don't have a shared stroage or to keep the migration time short. And for Disaster Recovery

Yes. You can also use it for HA, but with limitations.

• Sharad Storage to keep all VM Data into a Backup for Restore and else

Shared storage and backup are independent of each other. For HA, you put your guest's images onto the shared storage. Then each node can access the guest's disks even when a node goes down.

• HA is practical but only if you can guarantee the redundancy of the hardware. What can cause more costs

You need to have enough resources available such that guests can be recovered to different nodes in case of a node failure (e.g. if you have 3 nodes and 4 guests per node, each node should have enough resources to handle an additional 2 guests).

 

[croaky]

Thank you Fabian_E i was able to gain a better understanding through you