Github user liyezhang556520 commented on the pull request:
https://github.com/apache/spark/pull/3629#issuecomment-69684345
Hi @andrewor14 , I think @suyanNone 's concern is for code on
[line#290](https://github.com/apache/spark/blob/4e1f12d997426560226648d62ee17c90352613e7/core/src/main/scala/org/apache/spark/storage/MemoryStore.scala#L290),
which is `val amountToRelease = currentUnrollMemoryForThisThread -
previousMemoryReserved`. Since there is an impact from
`previousMemoryReserved`. We need to separate `unrollMap` and
`pendingUnrollMap`. But from my point of view, the value of
`previousMemoryReserved` should be always 0. @andrewor14 , can you tell in
which case `previousmemoryReserved` is supposed to have value? Because in my
opinion, each time we call `unrollSafely`, the currently task should either
have finished its previous memory process or it just newly start to process
it's first cache job.
---
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 [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
