[CORE] Decrease offheap memory size in resource profile for whole stage fallback case#8911
Conversation
|
Thanks for opening a pull request! Could you open an issue for this pull request on Github Issues? https://github.com/apache/incubator-gluten/issues Then could you also rename commit message and pull request title in the following format? See also: |
|
Run Gluten Clickhouse CI on x86 |
philo-he
force-pushed
the
decreass-offheap
branch
from
6c1c8c9 to
7b90f66
Compare
|
Run Gluten Clickhouse CI on x86 |
| executorResource.put(ResourceProfile.MEMORY, newExecutorMemory) | ||
|
|
||
| val newExecutorOffheap = | ||
| new ExecutorResourceRequest(ResourceProfile.OFFHEAP_MEM, offheapRequest.get.amount / 10) |
There was a problem hiding this comment.
@zjuwangg, could you take a look? This is just an empirical setting. And I am not sure which setting is better.
There was a problem hiding this comment.
LGTM in whole stage fallback case!
There was a problem hiding this comment.
This would work only for Spark 3.5.4 or higher as apache/spark#48963
There was a problem hiding this comment.
@philo-he Should we consider making this setting configurable?
There was a problem hiding this comment.
@JkSelf, let's first use this empirical fixed setting and wait for feedback from user side. In theory, user don't need to configure too much offheap memory in stage fallback case. Thanks!
There was a problem hiding this comment.
@philo-he
Thanks for improving on this!
just to confirm, I can see the code logic is to reduce the offheap memory size to 1/10, but on-heap memory size is not increased, is this intended?
Thanks, -yuan
There was a problem hiding this comment.
@zhouyuan, the on-heap memory is increased by specifying a larger value for ResourceProfile.MEMORY prior to the off-heap adjustment. If I have missed something, please let me know. Thanks!
| executorResource.put(ResourceProfile.MEMORY, newExecutorMemory) | ||
|
|
||
| val newExecutorOffheap = | ||
| new ExecutorResourceRequest(ResourceProfile.OFFHEAP_MEM, offheapRequest.get.amount / 10) |
There was a problem hiding this comment.
LGTM in whole stage fallback case!
philo-he
force-pushed
the
decreass-offheap
branch
from
7b90f66 to
3dde772
Compare
|
Run Gluten Clickhouse CI on x86 |
5 participants
What changes were proposed in this pull request?
For whole stage fallback case, we should not request the same offheap memory size as offload case.
How was this patch tested?
Fix a feature that is still experimental.