I have a similar challenge. My intended solution is to only use the ‘In Service’ upgrade, and not the ‘Rolling Service’ upgrade so that id will not change. However I am very nervous about that and figure it will likely change at some point. I also plan to run regular database dumps as a normal backup process, and possibly some kind of clustering between containers running at different sites. Oh and one other solution I may use is rancher-nfs mount: Is it possible to directly map rancher-nfs subfolders as volumes?