Skip to content

CNDB-14077: Reduce compaction thread pool size to match num of physical cores #1736

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
wants to merge 6 commits into
base: main
Choose a base branch
from
Open

CNDB-14077: Reduce compaction thread pool size to match num of physical cores #1736

Changes from 3 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
812e7ab
CNDB-14077: Reduce compaction thread pool size to match num of physic…
michaeljmarshall May 15, 2025
a15ceca
CNDB-14160: Fix IndexContext#getReferencedView (#1744)
michaeljmarshall May 20, 2025
6c96b35
Revert "CNDB-14160: Fix IndexContext#getReferencedView (#1744)"
michaeljmarshall May 23, 2025
9eb46e7
Add config: cassandra.sai.compaction.executor.threads
michaeljmarshall Jun 12, 2025
139a500
Temp: add metrics (will develop better metrics later)
michaeljmarshall Jun 12, 2025
543564d
Fix threadsafety issue in metrics
michaeljmarshall Jun 13, 2025
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
6 changes: 4 additions & 2 deletions src/java/org/apache/cassandra/index/sai/disk/v1/SegmentBuilder.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -78,8 +78,10 @@ public abstract class SegmentBuilder
{
private static final Logger logger = LoggerFactory.getLogger(SegmentBuilder.class);

/** for parallelism within a single compaction */
public static final ExecutorService compactionExecutor = new DebuggableThreadPoolExecutor(Runtime.getRuntime().availableProcessors(),
/** for parallelism within a single compaction
* see comments to JVector PhysicalCoreExecutor -- HT tends to cause contention for the SIMD units
*/
public static final ExecutorService compactionExecutor = new DebuggableThreadPoolExecutor(Runtime.getRuntime().availableProcessors() / 2,
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

what about making this configurable ? in case we need to rollback

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

fwiw, I copied this logic from:

cassandra/src/java/org/apache/cassandra/index/sai/disk/vector/CompactionGraph.java

Lines 104 to 108 in 61f57a6

// see comments to JVector PhysicalCoreExecutor -- HT tends to cause contention for the SIMD units
private static final ForkJoinPool compactionSimdPool = new ForkJoinPool(Runtime.getRuntime().availableProcessors() / 2,
new LowPriorityThreadFactory(),
null,
false);

Are we looking to make all of these configurable?

1,
TimeUnit.MINUTES,
new ArrayBlockingQueue<>(10 * Runtime.getRuntime().availableProcessors()),
Expand Down