It's C: batch_size is the number of instances that can be updated at a given time.
There's no direct relation between batch size and total number of shards/instances. E.g. for a job with 100 instances and a batch size of 10, at most 10 instances will be updating at a given time. If it turns out that all instances finish updating at exactly the same time, then it would take 10 batches to complete the update. However, the more likely scenario is that due to scheduling delays, startup times, etc., updates from the initial batch will complete at staggered intervals. As individual instances complete their update, additional instances will begin their updates to keep the number of instances currently updating at batch_size. Does that make sense? On Mon, Oct 30, 2017 at 4:01 PM, Mohit Jaggi <[email protected]> wrote: > Folks, > Does the following doc mean A or B? > > *A*: batch_size is the number of instances in a given shard > *B:* batch_size is the number of shards. So every batch has (number of > instances)/(batch_size) tasks. > > Mohit. > UpdateConfig Objects > > Parameters for controlling the rate and policy of rolling updates. > objecttypedescription > batch_size Integer Maximum number of shards to be updated in one > iteration (Default: 1) >
