Github user jinxing64 commented on the issue:
https://github.com/apache/spark/pull/18388
@tgravescs
Thanks a lot for advice.
> the flow control part should allow everyone to start fetching without
rejecting a bunch, especially if the network can't push it out that fast anyway.
For instance only create a handful of those outgoing buffers and wait to
get successfully sent messages back for the those before creating more.
Regarding flow control, can I ask a question?
In flow control, we won't reject `OpenBlocks`, but do we reject
`ChunkFetchRequest`? If we don't reject `ChunkFetchRequest`, it means we will
accept all `ChunkFetchRequest`s. We must store the `ChunkFetchRequest`s in some
buffer when memory pressure from Netty is above water mark. But buffering the
`ChunkFetchRequest`s can also take much memory. That will be another issue. The
problem is the client never know when to slow down sending requests.
Am I wrong?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
