JohnB agree, without knowing of site it is impossible to make complete no downtime transfer. By no downtime I rather meant instant switching to another backend.
As for moving website by itself, this can be done, for example, like this:
- First rsync, may take long but will transfer biggest part of data
- Second rsync, to transfer files that may be changed during first long rsync and make site most actual before switching
- Switching backend
- Final rsync to catch eventual changes during second one
Sure in case of sites which are constantly updating, some freeze option will be needed after step 2, until step 4 is finished. But anyway, for really big structures instant move is impossible.
JohnB a feature where of you move a site to a new server the old server will proxy requests until DNS propagation which I think is a good medium.
Well, we run here into problems too:
- How the panel will detect if external DNS caches were updated and it is safe to remove proxy from old server?
- If customer is using not ours DNS, we need to ask him to change IP. But before that, we will need to check in whois, which nameservers are set for every moved domain
P.S. I wonder why so many guys here are against this feature. It will make life much easier, like, for example, live migration of virtual machines from node to node in Proxmox cluster - there is no need to warn user, stop VM, watch over migration and start it again - you just press button and it getting moved: without reboot, with all daemons running and without your care. As I see, Enhance is a cluster-targeted panel and exactly this makes it unique. So to build a cluster, easy moving from one node to another should just exist.
