Qinlong Wang
Qinlong Wang
@jjyao Can you share how to solve it?
解释下使用 cachedJob 的原因是,job 里记录了pod 的状态信息。为了避免每次都访问 apiserver 来获取 job 下面的 pod 状态,所以就在监听到 pod event 的时候把 pod 状态记录到 job 对象里。
Fix in #1552
Yes, I’ll submit the refactor design of master soon. But, we need to discuss whether to keep the brain which is not commonly used.
Thanks for your suggestion. Can we mitigate this risk if we replace GRPC and pickle with HTTP and JSON? IMO, JSON is more safe than pickle.