Good and simple way to use replication ?

[amianalien0x3f]

Hello
I'm quite familiar with Hyper-V replication and migration and I don't really understand the way replication has to be used with proxmox
My need is I just want to have 2 nodes on 2 physical hosts and use replication to be able to start the replicated VM on the second physical Host as soom as the first Host failed. This can be manually, my goald is that it has to be simple for user, when node 1 fails, just start node 2
I have seen that replication creates a copy of the disk on the second node, but unlike Hyper-V, it seems to copy only the storage not the whole VM.
If the second Host fails, I can no longer use migration and I don't know what to do with the replicated disk. Do I have to create a second rescue VM on the second host using the replicated disk on the second host ?
I've tried but the replicated disk is not proposed in the list of disk available when creating this second VM on the second node...
So I think I miss something in the way replication is managed and used with proxmox
My other question is the way to use replication source-targer. I've seen that replication is correctly and automatically updated and swapped when you migrate a VM from one node to another (node1=>node2 then autonmatically node2=>node1 after migration) but is as simple when first host fails totally and the migration systems is so unavailable, but only replicated VM ?
Thank you for your explanations in advance
RV

 

[apoc]

First thing to mention is that KVM has a different concept than Hyper-V and also ESXi.
So applying concepts from one to the other doesn't fit. Its just like that...

Putting that aside the disk is basically what matters most. So from that perspective you are fine for a DR purpose.

A migration can't be done if one host is unavailable. That sounds logical to me. So in this case you would likely need to rebuild configuration of the VM on the DR site.
Its good practice anyways to backup the VM configuration files from /etc/pve/qemu-server directory. I am copying them down to a separate place regularily and I'd go and check if those config files do the trick for you.

 

[DerDanilo]

First thing to mention is that KVM has a different concept than Hyper-V and also ESXi.
So applying concepts from one to the other doesn't fit. Its just like that...

Putting that aside the disk is basically what matters most. So from that perspective you are fine for a DR purpose.

A migration can't be done if one host is unavailable. That sounds logical to me. So in this case you would likely need to rebuild configuration of the VM on the DR site.
Its good practice anyways to backup the VM configuration files from /etc/pve/qemu-server directory. I am copying them down to a separate place regularily and I'd go and check if those config files do the trick for you.

If the nodes run in a cluster, which is usually required when using the buildin replication, it is fairly simple by just moving the VM configuration file to the corresponding folder of the node it then should be running on. (/etc/pve/nodes/...)
Don't know if it is meanwhile supported in the UI/api to migrate offline VMs to another node, from a dead node.

 

[amianalien0x3f]

Thank you Tburger and DerDanilo
I have the feeling I'm missing something of how to deal with replication with proxmox. I don't see the use of replicationin in fact
My aim is not to launch a big debate, but replication every 15 minutes doesn't protect from data corruption. You must have an incremental backup for that on a certain period. So having a fast instant replication allows to protect you from physical failure. So the first thing you want when an host fails and when you have a backup somewhere is to launch the backup on the other host as soon as possible (automatically or not is another debate) . It seems that is not planned like that and quite so simple to do that with proxmox actually.
So what the use of replication ?
Another question is what to do with the replicated storage so ?
I think I will try to look for a solution based on VM files replication as you mentionned both, but as it seems not to be designed to do so. My fear is to create other problems. First of all the unkown behaviour when both VM (one original, the second using the replicated) are launch at the same time by accident.
Second one is the management of inversion of replication.
Third maybe I'm totally wrong of the use of replication system and what I want to do ^^

 

[hostedio]

I'm sorry to the portrayal of the subject again. I'm looking for answers to the same questions.

i have 2 physical machines. the cluster is included and replication is working. But 2. i see only the data of the replication on the machine called "vm-100-disk-0". 2. Use this data. i didn't understand how to stand up on the machine.


The operation I want to do is 1. 2. the virtual machine on the machine. i send a copy to the machine every day and guarantee myself.