-
Notifications
You must be signed in to change notification settings - Fork 42
Open
Description
| a.requestChunk(chunkIdx) |
When chunk hasn't arrived from server yet, the consumer code that is calling Read() will get 0-byte successful reads at rapid pace, until server responds. This means that the consumer is calling the read as fast as it can, millions of times a second, manifesting in 100 % CPU usage. After I inserted a 100 ms sleep in after the line I quoted, CPU usage is back to normal.
This behaviour is discouraged by the Reader interface:
Implementations of Read are discouraged from returning a zero byte count with a nil error
It doesn't say it directly, but it means that Read() is expected to block until it has data to return.
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
No labels