Glance Caching sleep time should probably be a configurable option

[carpnick]

This line should be configurable.

Linux images are fine with that requirement. But Windows images, we have seen go much higher with 15-18 GB images.

[jjasghar]

wait for server doesn't do what you want?

[carpnick]

Figured that was for a different use case. If it's the same use case, maybe we should consider removing the random 30 second wait? Then default the wait_for_server to be 30 via configurable option?

[jjasghar]

Nope, the "sleep 30" is due to a caching issue that glance has. Yep, it's gross, but it's the only way we've been able to figure out how to get around it.

It's literally called glance_cache_something or something. I'm not recalling the name at the moment unfortunately.

[carpnick]

Right so the cache issue is due to image download to the DS because its not in the cache, correct? Basically what I am wondering is, what if we have a bigger image? Your first comment appeared to say use wait_for_server, but this call happens after the floating ip call.

[carpnick]

Another option that we implemented, that we could do a PR, we waited for the VM state to change to ready.

Something like this: server.wait_for(600) { print ".";ready? }

[jjasghar]

I've put the PR in, please take a look, suggestions are more then welcome.