I have been trying to find the answer to my question below but I couldn't so I'm posting the question here. Not sure if it's been asked before. Sorry in advance if it has, and in that case please point me to it.
Currently Glu only supports serial and parallel execution plan. If I, say, have 10 entries for 10 different agents/machines. Now I want to perform an action (e.g. deploy, bounce). If choose serial execution plan, then Glu will perform the action one action/machine at a time. And if choose parallel execution, then Glu will perform the action on all 10 agents/machines at once.
However, is there a hybrid option that I can have Glu perform the action in parallel on N (say 2) number of agents/machines, then move on to another N again in parallel, until the action have been performed on all agents/machines?
I actually like your idea/concept of the scale of hybrid plan: on one end of the scale it is fully parallel, on the other end of the scale it is fully sequential. I am just wondering what would be the best way to visually represent it (parallel and sequential are easy... hybrid requires an extra parameter N)... Any suggestion would be appreciated (created a ticket for this feature: https://github.com/pongasoft/glu/issues/260)
Thank you Yan for the information. We in fact completely control Glu via REST and build lots of abstractions on top it to generate proper static models. However, giving the complexity and the fact that it's not a showstopper for us right now, I'll wait until this feature comes out. Some apps of our have lots of hosts. This feature will significantly speed up the entire process.