Hi, I tried this: in an running topology, I killed a worker (by doing "kill" command to the worker progress), then I found that it did restart the worker, which however took a "staring" time. It seems not a hot backup. Is there something like "worker swapping"? As for the "swap" function, I read about it's on the roadmap from Google Groups. So what is the exact function - I mean if just one worker gets down, is the whole topology swapped? And is the "swap" function available?
PS: I have asked the same question about "swap" in another mail "storm swap available? ", please ignore that mail. Best regards, Hao Yu
