What I’ve noticed so far is that downloading the ZOS v2 takes even longer than on v1 and when starting Bancadati/tlre I need to be really careful not to cause download timeouts.
Since it downloads the image over HTTPS this already eliminates a lot of caching options, unless I would be able to do an “ssl/tls downgrade attack” which is not something I would be comfortable with.
I still need to check if downloads would skip TLS verification, in that case I could still let bootstrap.grid.tf point to a local cache server. I still would not be 100% comfortable if this was the case however.
I’ve seen that there are multiple farms that have quite a few nodes in them. Assuming that most of those are in the same area and even network, an easy caching solution (e.g.: deploy a docker container and tweak some DNS settings) would eliminate a lot of bandwidth usage, have much better boot times and many other advantages.
If this could be expanded to most/all downloadable components of ZoS, that would be pretty cool!
Not the most urgent thing as the nodes are expected to stay on and don’t need a reboot for updates anymore (correct me if I’m wrong here).
However, it would make deploying a new farm or restarting a farm after a breakdown, e.g. power outage a lot quicker and smoother.
Food for thought.