Skip to content

[SPARK-57242][CORE] Avoid unbounded page allocation retries after allocator OOM #56293

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
sunchao wants to merge 10 commits into
base: master
Choose a base branch
from
Open

[SPARK-57242][CORE] Avoid unbounded page allocation retries after allocator OOM #56293

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
d25673e
[SPARK-57242][CORE] Avoid unbounded page allocation retries after all…
sunchao Jun 3, 2026
3999335
[SPARK-57242][CORE] Fix shuffle pointer state and spill accounting
sunchao Jun 3, 2026
3fe46a8
[SPARK-57242][CORE] Preserve usable shuffle pointer array after spill
sunchao Jun 3, 2026
b521283
[SPARK-57242][CORE] Preserve bounded partial page recovery
sunchao Jun 3, 2026
e2cde21
[SPARK-57242][CORE] Combine spill recovery with free-tail memory
sunchao Jun 3, 2026
7c1f650
[SPARK-57242][CORE] Restore minimum shuffle pointer array after spill
sunchao Jun 4, 2026
7dc2c94
[SPARK-57242][CORE] Clarify spill progress accounting
sunchao Jun 4, 2026
f4bcd7d
[SPARK-57242][CORE] Preserve shuffle pointer growth after spill
sunchao Jun 5, 2026
4a1cee9
update
sunchao Jun 15, 2026
d08d28d
More update
sunchao Jun 17, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 10 additions & 2 deletions core/src/main/java/org/apache/spark/memory/MemoryConsumer.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -73,7 +73,9 @@ public void spill() throws IOException {
*
* This should be implemented by subclass.
*
* Note: In order to avoid possible deadlock, should not call acquireMemory() from spill().
* Note: In order to avoid possible deadlock, implementations must release memory synchronously
* on the calling thread and must not acquire task memory from spill(), either directly or from
* another thread.
*
* Note: today, this only frees Tungsten-managed pages.
*
Expand Down Expand Up @@ -115,7 +117,8 @@ public void freeArray(LongArray array) {
* @throws SparkOutOfMemoryError
*/
protected MemoryBlock allocatePage(long required) {
MemoryBlock page = taskMemoryManager.allocatePage(Math.max(pageSize, required), this);
MemoryBlock page =
taskMemoryManager.allocatePageWithMinimum(Math.max(pageSize, required), required, this);
if (page == null || page.size() < required) {
throwOom(page, required);
}
Expand All @@ -131,6 +134,11 @@ protected void freePage(MemoryBlock page) {
taskMemoryManager.freePage(page, this);
}

/** Returns whether this page came from a minimum retry after a partial allocation failed. */
protected boolean isPageAllocationFromMinimumRetry(MemoryBlock page) {
return taskMemoryManager.isPageAllocationFromMinimumRetry(page);
}

/**
* Allocates memory of `size`.
*/
Expand Down
Loading