HA, Replication and iSCSI.

[freebee]

Hi.
I have this scenario 1:

1. A ISCSI server give the Harddisk to every VM.
2. 6 Hosts is used in HA, and all see this iSCSI server.
3. I want to replicate each minute this VM to outside building. For this i have a VPN and is working (the 6 hosts + iSCSI Server see my other place network and my network see the proxmox hosts network).
4. The only way proxmox accept to send replication is if i add "other" proxmox to the cluster. Ok.
6. Can i install the other proxmox host here and add to the cluster ?. The ISCSI server accept connections from here.
7. If yes, this replication will be stored on iSCSI server or in my localstorage outside the building ?.

I see about the proxmox backup server and maybe is a solution, however, i don't see how "make backup" every minute, unless i make some bash script loop (if the backup-client can be used from console). If someone have more info about proxmox backup server in this context will be welcome suggestions.

 

[aaron]

I don't think you can make this work as you expect. The PVE replication requires ZFS as underlying storage. ZFS can not be used as shared storage directly.

What is the iSCSI server? Can you implement the replication there?

 

[freebee]

I don't think you can make this work as you expect. The PVE replication requires ZFS as underlying storage. ZFS can not be used as shared storage directly.

What is the iSCSI server? Can you implement the replication there?

Hi.
I use the FreeBSD ZFS. But, i changing the the view where storage is local, with low complexity. In this new view, the servers is zfs and replication is from hypervisor to other.

 

[guletz]

Hi,

The zfs replication is safe if your source and the destination are using the same OS on both ends. Sometimes are some zfs bugs/problem who are patched by PMX devs. I am not sure that the same patch are apply on the same time to others OS.

And I can also told that for several years I must have to run various serviceces in location A and location B with some redundancy (fail-over mostly). Sometimes this services has not work as expected (different OSes implementation). After seveseveral fails I decided to use the same OS(I had in use slackware and fedora) and the same version/updates.

With this decision I was able to solve my past problems. And also my admin tasks was less time consuming. Also the new hardware decisions to buy new components.

In conclusion for me, the best way is to use the same hardware/sofware in both locations.


Good luck / Bafta !

 

[freebee]

Hi.
Thank for you feedback. Is this i trying to archive: one OS to operate everything.
I making this test:
watch -n 60 vzdump 300610 --mode snapshot --storage DC2-GN-HD01 --remove 0 --node PV-01-GN
Every 60 seconds i send the differential to the backup-proxmox server.
Any chance to got node unstable if use like that ?.

 

[guletz]

Any chance to got node unstable if use like that ?.

Here the only problem will be the load/IO pressure made by pve-zsync.