You are viewing a plain text version of this content. The canonical link for it is here.
Posted to pr@cassandra.apache.org by "blambov (via GitHub)" <gi...@apache.org> on 2024/11/19 10:53:21 UTC
[PR] Introduce SSTableSimpleScanner for compaction [cassandra]
blambov opened a new pull request, #3695:
URL: https://github.com/apache/cassandra/pull/3695
Thanks for sending a pull request! Here are some tips if you're new here:
* Ensure you have added or run the [appropriate tests](https://cassandra.apache.org/_/development/testing.html) for your PR.
* Be sure to keep the PR description updated to reflect all changes.
* Write your PR title to summarize what this PR proposes.
* If possible, provide a concise example to reproduce the issue for a faster review.
* Read our [contributor guidelines](https://cassandra.apache.org/_/development/index.html)
* If you're making a documentation change, see our [guide to documentation contribution](https://cassandra.apache.org/_/development/documentation.html)
Commit messages should follow the following format:
```
<One sentence description, usually Jira title or CHANGES.txt summary>
<Optional lengthier description (context on patch)>
patch by <Authors>; reviewed by <Reviewers> for CASSANDRA-#####
Co-authored-by: Name1 <email1>
Co-authored-by: Name2 <email2>
```
The [Cassandra Jira](https://issues.apache.org/jira/projects/CASSANDRA/issues/)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: pr-unsubscribe@cassandra.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: pr-unsubscribe@cassandra.apache.org
For additional commands, e-mail: pr-help@cassandra.apache.org
Re: [PR] CASSANDRA-20092: Introduce SSTableSimpleScanner for compaction [cassandra]
Posted by "pcmanus (via GitHub)" <gi...@apache.org>.
pcmanus commented on code in PR #3695:
URL: https://github.com/apache/cassandra/pull/3695#discussion_r1865658640
##########
src/java/org/apache/cassandra/io/sstable/format/SSTableReader.java:
##########
@@ -728,27 +728,104 @@ public List<PartitionPositionBounds> getPositionsForRanges(Collection<Range<Toke
for (Range<Token> range : Range.normalize(ranges))
{
assert !range.isWrapAround() || range.right.isMinimum();
- // truncate the range so it at most covers the sstable
AbstractBounds<PartitionPosition> bounds = Range.makeRowRange(range);
- PartitionPosition leftBound = bounds.left.compareTo(first) > 0 ? bounds.left : first.getToken().minKeyBound();
- PartitionPosition rightBound = bounds.right.isMinimum() ? last.getToken().maxKeyBound() : bounds.right;
+ PartitionPositionBounds pb = getPositionsForBounds(bounds);
+ if (pb != null)
+ positions.add(pb);
+ }
+ return positions;
+ }
- if (leftBound.compareTo(last) > 0 || rightBound.compareTo(first) < 0)
- continue;
+ /**
+ * Get a list of data positions in this SSTable that correspond to the given list of bounds. This method will remove
+ * non-covered intervals, but will not correct order or overlap in the supplied list, e.g. if bounds overlap, the
+ * result will be sections of the data file that repeat the same positions.
+ *
+ * @return A sorted list of [offset,end) pairs corresponding to the given boundsList in the datafile for this
+ * SSTable.
+ */
+ public List<PartitionPositionBounds> getPositionsForBoundsIterator(Iterator<AbstractBounds<PartitionPosition>> boundsList)
+ {
+ // use the index to determine a minimal section for each range
+ List<PartitionPositionBounds> positions = new ArrayList<>();
+ while (boundsList.hasNext())
+ {
+ AbstractBounds<PartitionPosition> bounds = boundsList.next();
+ PartitionPositionBounds pb = getPositionsForBounds(bounds);
+ if (pb != null)
+ positions.add(pb);
+ }
+ return positions;
+ }
- long left = getPosition(leftBound, Operator.GT);
- long right = (rightBound.compareTo(last) > 0)
- ? uncompressedLength()
- : getPosition(rightBound, Operator.GT);
+ /**
+ * Determine the data positions in this SSTable that cover the given bounds.
+ *
+ * @return An [offset,end) pair that cover the given bounds in the datafile for this SSTable.
Review Comment:
We should specify that this can return `null`, and under which conditions.
##########
src/java/org/apache/cassandra/io/sstable/format/SSTableReader.java:
##########
@@ -728,27 +728,104 @@ public List<PartitionPositionBounds> getPositionsForRanges(Collection<Range<Toke
for (Range<Token> range : Range.normalize(ranges))
{
assert !range.isWrapAround() || range.right.isMinimum();
- // truncate the range so it at most covers the sstable
AbstractBounds<PartitionPosition> bounds = Range.makeRowRange(range);
- PartitionPosition leftBound = bounds.left.compareTo(first) > 0 ? bounds.left : first.getToken().minKeyBound();
- PartitionPosition rightBound = bounds.right.isMinimum() ? last.getToken().maxKeyBound() : bounds.right;
+ PartitionPositionBounds pb = getPositionsForBounds(bounds);
+ if (pb != null)
+ positions.add(pb);
+ }
+ return positions;
+ }
- if (leftBound.compareTo(last) > 0 || rightBound.compareTo(first) < 0)
- continue;
+ /**
+ * Get a list of data positions in this SSTable that correspond to the given list of bounds. This method will remove
+ * non-covered intervals, but will not correct order or overlap in the supplied list, e.g. if bounds overlap, the
+ * result will be sections of the data file that repeat the same positions.
+ *
+ * @return A sorted list of [offset,end) pairs corresponding to the given boundsList in the datafile for this
+ * SSTable.
+ */
+ public List<PartitionPositionBounds> getPositionsForBoundsIterator(Iterator<AbstractBounds<PartitionPosition>> boundsList)
+ {
+ // use the index to determine a minimal section for each range
+ List<PartitionPositionBounds> positions = new ArrayList<>();
+ while (boundsList.hasNext())
+ {
+ AbstractBounds<PartitionPosition> bounds = boundsList.next();
+ PartitionPositionBounds pb = getPositionsForBounds(bounds);
+ if (pb != null)
+ positions.add(pb);
+ }
+ return positions;
+ }
- long left = getPosition(leftBound, Operator.GT);
- long right = (rightBound.compareTo(last) > 0)
- ? uncompressedLength()
- : getPosition(rightBound, Operator.GT);
+ /**
+ * Determine the data positions in this SSTable that cover the given bounds.
+ *
+ * @return An [offset,end) pair that cover the given bounds in the datafile for this SSTable.
+ */
+ public PartitionPositionBounds getPositionsForBounds(AbstractBounds<PartitionPosition> bounds)
+ {
+ long left = getPosition(bounds.left, bounds.inclusiveLeft() ? Operator.GE : Operator.GT);
+ // Note: getPosition will apply a moved start if the sstable is in MOVED_START state.
+ if (left < 0) // empty range
+ return null;
- if (left == right)
- // empty range
- continue;
+ long right = bounds.right.isMinimum() ? -1
+ : getPosition(bounds.right, bounds.inclusiveRight() ? Operator.GT
+ : Operator.GE);
+ if (right < 0) // right is beyond end
+ right = uncompressedLength(); // this should also be correct for EARLY readers
- assert left < right : String.format("Range=%s openReason=%s first=%s last=%s left=%d right=%d", range, openReason, first, last, left, right);
- positions.add(new PartitionPositionBounds(left, right));
+ if (left >= right)
+ {
+ // empty range
+ return null;
+ }
+
+ return new PartitionPositionBounds(left, right);
+ }
+
+ /**
+ * Return a (offset, end] pair that covers the whole file.
+ */
+ public PartitionPositionBounds getPositionsForFullRange()
+ {
+ if (openReason != OpenReason.MOVED_START)
+ return new PartitionPositionBounds(0, uncompressedLength());
+ else
+ {
+ // query a full range, so that the required adjustments can be applied
+ PartitionPosition minToken = getPartitioner().getMinimumToken().minKeyBound();
+ return getPositionsForBounds(new Range<>(minToken, minToken));
}
- return positions;
+ }
+
+ /**
+ * Calculate a total on-disk (compressed) size for the given partition positions.
Review Comment:
Maybe worth specifying:
1. what it returns exactly in the case of compression. Namely that it's the size of the chunks covering the positions, and so technically can include more than the requested bounds and how much is subject to the chunk size (yes, it's unclear how this could be anything else if you think about it, but still worth calling out imo).
2. that this assume the provided bounds are non-overlapping.
##########
src/java/org/apache/cassandra/io/sstable/format/SSTableReader.java:
##########
@@ -728,27 +728,104 @@ public List<PartitionPositionBounds> getPositionsForRanges(Collection<Range<Toke
for (Range<Token> range : Range.normalize(ranges))
{
assert !range.isWrapAround() || range.right.isMinimum();
- // truncate the range so it at most covers the sstable
AbstractBounds<PartitionPosition> bounds = Range.makeRowRange(range);
- PartitionPosition leftBound = bounds.left.compareTo(first) > 0 ? bounds.left : first.getToken().minKeyBound();
- PartitionPosition rightBound = bounds.right.isMinimum() ? last.getToken().maxKeyBound() : bounds.right;
+ PartitionPositionBounds pb = getPositionsForBounds(bounds);
+ if (pb != null)
+ positions.add(pb);
+ }
+ return positions;
+ }
- if (leftBound.compareTo(last) > 0 || rightBound.compareTo(first) < 0)
- continue;
+ /**
+ * Get a list of data positions in this SSTable that correspond to the given list of bounds. This method will remove
+ * non-covered intervals, but will not correct order or overlap in the supplied list, e.g. if bounds overlap, the
+ * result will be sections of the data file that repeat the same positions.
+ *
+ * @return A sorted list of [offset,end) pairs corresponding to the given boundsList in the datafile for this
+ * SSTable.
+ */
+ public List<PartitionPositionBounds> getPositionsForBoundsIterator(Iterator<AbstractBounds<PartitionPosition>> boundsList)
+ {
+ // use the index to determine a minimal section for each range
+ List<PartitionPositionBounds> positions = new ArrayList<>();
+ while (boundsList.hasNext())
+ {
+ AbstractBounds<PartitionPosition> bounds = boundsList.next();
+ PartitionPositionBounds pb = getPositionsForBounds(bounds);
+ if (pb != null)
+ positions.add(pb);
+ }
+ return positions;
+ }
- long left = getPosition(leftBound, Operator.GT);
- long right = (rightBound.compareTo(last) > 0)
- ? uncompressedLength()
- : getPosition(rightBound, Operator.GT);
+ /**
+ * Determine the data positions in this SSTable that cover the given bounds.
+ *
+ * @return An [offset,end) pair that cover the given bounds in the datafile for this SSTable.
+ */
+ public PartitionPositionBounds getPositionsForBounds(AbstractBounds<PartitionPosition> bounds)
+ {
+ long left = getPosition(bounds.left, bounds.inclusiveLeft() ? Operator.GE : Operator.GT);
+ // Note: getPosition will apply a moved start if the sstable is in MOVED_START state.
+ if (left < 0) // empty range
+ return null;
- if (left == right)
- // empty range
- continue;
+ long right = bounds.right.isMinimum() ? -1
+ : getPosition(bounds.right, bounds.inclusiveRight() ? Operator.GT
+ : Operator.GE);
+ if (right < 0) // right is beyond end
+ right = uncompressedLength(); // this should also be correct for EARLY readers
- assert left < right : String.format("Range=%s openReason=%s first=%s last=%s left=%d right=%d", range, openReason, first, last, left, right);
- positions.add(new PartitionPositionBounds(left, right));
+ if (left >= right)
+ {
+ // empty range
+ return null;
+ }
+
+ return new PartitionPositionBounds(left, right);
+ }
+
+ /**
+ * Return a (offset, end] pair that covers the whole file.
+ */
+ public PartitionPositionBounds getPositionsForFullRange()
+ {
+ if (openReason != OpenReason.MOVED_START)
+ return new PartitionPositionBounds(0, uncompressedLength());
+ else
+ {
+ // query a full range, so that the required adjustments can be applied
+ PartitionPosition minToken = getPartitioner().getMinimumToken().minKeyBound();
+ return getPositionsForBounds(new Range<>(minToken, minToken));
Review Comment:
I'd add an assertion that the return of `getPositionsForBounds` can never be `null` here (since the method "can" return `null`, just not in that case (or at least I think; I'm not that familiar with `OpenReason.MOVED_START` but I assume we never move it to a point where the sstable is essentially empty?).
##########
src/java/org/apache/cassandra/io/sstable/format/SSTableSimpleScanner.java:
##########
@@ -0,0 +1,185 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.cassandra.io.sstable.format;
+
+import java.util.Collection;
+import java.util.Iterator;
+import java.util.NoSuchElementException;
+import java.util.Set;
+import java.util.concurrent.atomic.AtomicBoolean;
+
+import com.google.common.collect.ImmutableSet;
+
+import org.apache.cassandra.db.DecoratedKey;
+import org.apache.cassandra.db.rows.UnfilteredRowIterator;
+import org.apache.cassandra.io.sstable.CorruptSSTableException;
+import org.apache.cassandra.io.sstable.ISSTableScanner;
+import org.apache.cassandra.io.sstable.SSTableIdentityIterator;
+import org.apache.cassandra.io.util.RandomAccessReader;
+import org.apache.cassandra.schema.TableMetadata;
+
+import static org.apache.cassandra.io.sstable.format.SSTableReader.PartitionPositionBounds;
+
+/// Simple SSTable scanner that reads sequentially through an SSTable without using the index.
Review Comment:
Nit: are we sure the project is ok with using this new style of javadoc? I'm personally fine with it, but this imply someone interested by generating the javadoc would now need to use a fairly recent java (a technically unsupported by the project one) to get it javadoc-ified (which could slightly impact IDE that aren't up to date on this too).
##########
src/java/org/apache/cassandra/io/sstable/format/SSTableSimpleScanner.java:
##########
@@ -0,0 +1,185 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.cassandra.io.sstable.format;
+
+import java.util.Collection;
+import java.util.Iterator;
+import java.util.NoSuchElementException;
+import java.util.Set;
+import java.util.concurrent.atomic.AtomicBoolean;
+
+import com.google.common.collect.ImmutableSet;
+
+import org.apache.cassandra.db.DecoratedKey;
+import org.apache.cassandra.db.rows.UnfilteredRowIterator;
+import org.apache.cassandra.io.sstable.CorruptSSTableException;
+import org.apache.cassandra.io.sstable.ISSTableScanner;
+import org.apache.cassandra.io.sstable.SSTableIdentityIterator;
+import org.apache.cassandra.io.util.RandomAccessReader;
+import org.apache.cassandra.schema.TableMetadata;
+
+import static org.apache.cassandra.io.sstable.format.SSTableReader.PartitionPositionBounds;
+
+/// Simple SSTable scanner that reads sequentially through an SSTable without using the index.
+///
+/// This is a significant improvement for the performance of compaction over using the full-blown DataRange-capable
+/// [SSTableScanner] and enables correct calculation of data sizes to process.
+public class SSTableSimpleScanner
+implements ISSTableScanner
+{
+ private final AtomicBoolean isClosed = new AtomicBoolean(false);
+ private final RandomAccessReader dfile;
+ private final SSTableReader sstable;
+
+ private final Iterator<PartitionPositionBounds> rangeIterator;
+
+ private long bytesScannedInPreviousRanges;
+
+ private final long sizeInBytes;
+ private final long compressedSizeInBytes;
+
+ private long currentEndPosition;
+ private long currentStartPosition;
+
+ private SSTableIdentityIterator currentIterator;
+ private DecoratedKey lastKey;
+
+ public SSTableSimpleScanner(SSTableReader sstable,
Review Comment:
Maybe worth specifying that `boundList` must be non-overlapping.
##########
src/java/org/apache/cassandra/io/sstable/format/bti/BtiTableReader.java:
##########
@@ -382,27 +380,6 @@ public UnfilteredRowIterator rowIterator(FileDataInput dataFileInput,
return new SSTableIterator(this, dataFileInput, key, indexEntry, slices, selectedColumns, rowIndexFile);
}
- @Override
Review Comment:
This makes a bunch of the static `BtiScanner#getScanner` methods unused, and since we probably want those methods to never be used, they should probably be removed. Same for `BigTableScanner`.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: pr-unsubscribe@cassandra.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: pr-unsubscribe@cassandra.apache.org
For additional commands, e-mail: pr-help@cassandra.apache.org
Re: [PR] CASSANDRA-20092: Introduce SSTableSimpleScanner for compaction [cassandra]
Posted by "blambov (via GitHub)" <gi...@apache.org>.
blambov commented on code in PR #3695:
URL: https://github.com/apache/cassandra/pull/3695#discussion_r1867304834
##########
src/java/org/apache/cassandra/io/sstable/format/SSTableReader.java:
##########
@@ -728,27 +728,104 @@ public List<PartitionPositionBounds> getPositionsForRanges(Collection<Range<Toke
for (Range<Token> range : Range.normalize(ranges))
{
assert !range.isWrapAround() || range.right.isMinimum();
- // truncate the range so it at most covers the sstable
AbstractBounds<PartitionPosition> bounds = Range.makeRowRange(range);
- PartitionPosition leftBound = bounds.left.compareTo(first) > 0 ? bounds.left : first.getToken().minKeyBound();
- PartitionPosition rightBound = bounds.right.isMinimum() ? last.getToken().maxKeyBound() : bounds.right;
+ PartitionPositionBounds pb = getPositionsForBounds(bounds);
+ if (pb != null)
+ positions.add(pb);
+ }
+ return positions;
+ }
- if (leftBound.compareTo(last) > 0 || rightBound.compareTo(first) < 0)
- continue;
+ /**
+ * Get a list of data positions in this SSTable that correspond to the given list of bounds. This method will remove
+ * non-covered intervals, but will not correct order or overlap in the supplied list, e.g. if bounds overlap, the
+ * result will be sections of the data file that repeat the same positions.
+ *
+ * @return A sorted list of [offset,end) pairs corresponding to the given boundsList in the datafile for this
+ * SSTable.
+ */
+ public List<PartitionPositionBounds> getPositionsForBoundsIterator(Iterator<AbstractBounds<PartitionPosition>> boundsList)
+ {
+ // use the index to determine a minimal section for each range
+ List<PartitionPositionBounds> positions = new ArrayList<>();
+ while (boundsList.hasNext())
+ {
+ AbstractBounds<PartitionPosition> bounds = boundsList.next();
+ PartitionPositionBounds pb = getPositionsForBounds(bounds);
+ if (pb != null)
+ positions.add(pb);
+ }
+ return positions;
+ }
- long left = getPosition(leftBound, Operator.GT);
- long right = (rightBound.compareTo(last) > 0)
- ? uncompressedLength()
- : getPosition(rightBound, Operator.GT);
+ /**
+ * Determine the data positions in this SSTable that cover the given bounds.
+ *
+ * @return An [offset,end) pair that cover the given bounds in the datafile for this SSTable.
Review Comment:
Added.
##########
src/java/org/apache/cassandra/io/sstable/format/SSTableReader.java:
##########
@@ -728,27 +728,104 @@ public List<PartitionPositionBounds> getPositionsForRanges(Collection<Range<Toke
for (Range<Token> range : Range.normalize(ranges))
{
assert !range.isWrapAround() || range.right.isMinimum();
- // truncate the range so it at most covers the sstable
AbstractBounds<PartitionPosition> bounds = Range.makeRowRange(range);
- PartitionPosition leftBound = bounds.left.compareTo(first) > 0 ? bounds.left : first.getToken().minKeyBound();
- PartitionPosition rightBound = bounds.right.isMinimum() ? last.getToken().maxKeyBound() : bounds.right;
+ PartitionPositionBounds pb = getPositionsForBounds(bounds);
+ if (pb != null)
+ positions.add(pb);
+ }
+ return positions;
+ }
- if (leftBound.compareTo(last) > 0 || rightBound.compareTo(first) < 0)
- continue;
+ /**
+ * Get a list of data positions in this SSTable that correspond to the given list of bounds. This method will remove
+ * non-covered intervals, but will not correct order or overlap in the supplied list, e.g. if bounds overlap, the
+ * result will be sections of the data file that repeat the same positions.
+ *
+ * @return A sorted list of [offset,end) pairs corresponding to the given boundsList in the datafile for this
+ * SSTable.
+ */
+ public List<PartitionPositionBounds> getPositionsForBoundsIterator(Iterator<AbstractBounds<PartitionPosition>> boundsList)
+ {
+ // use the index to determine a minimal section for each range
+ List<PartitionPositionBounds> positions = new ArrayList<>();
+ while (boundsList.hasNext())
+ {
+ AbstractBounds<PartitionPosition> bounds = boundsList.next();
+ PartitionPositionBounds pb = getPositionsForBounds(bounds);
+ if (pb != null)
+ positions.add(pb);
+ }
+ return positions;
+ }
- long left = getPosition(leftBound, Operator.GT);
- long right = (rightBound.compareTo(last) > 0)
- ? uncompressedLength()
- : getPosition(rightBound, Operator.GT);
+ /**
+ * Determine the data positions in this SSTable that cover the given bounds.
+ *
+ * @return An [offset,end) pair that cover the given bounds in the datafile for this SSTable.
+ */
+ public PartitionPositionBounds getPositionsForBounds(AbstractBounds<PartitionPosition> bounds)
+ {
+ long left = getPosition(bounds.left, bounds.inclusiveLeft() ? Operator.GE : Operator.GT);
+ // Note: getPosition will apply a moved start if the sstable is in MOVED_START state.
+ if (left < 0) // empty range
+ return null;
- if (left == right)
- // empty range
- continue;
+ long right = bounds.right.isMinimum() ? -1
+ : getPosition(bounds.right, bounds.inclusiveRight() ? Operator.GT
+ : Operator.GE);
+ if (right < 0) // right is beyond end
+ right = uncompressedLength(); // this should also be correct for EARLY readers
- assert left < right : String.format("Range=%s openReason=%s first=%s last=%s left=%d right=%d", range, openReason, first, last, left, right);
- positions.add(new PartitionPositionBounds(left, right));
+ if (left >= right)
+ {
+ // empty range
+ return null;
+ }
+
+ return new PartitionPositionBounds(left, right);
+ }
+
+ /**
+ * Return a (offset, end] pair that covers the whole file.
+ */
+ public PartitionPositionBounds getPositionsForFullRange()
+ {
+ if (openReason != OpenReason.MOVED_START)
+ return new PartitionPositionBounds(0, uncompressedLength());
+ else
+ {
+ // query a full range, so that the required adjustments can be applied
+ PartitionPosition minToken = getPartitioner().getMinimumToken().minKeyBound();
+ return getPositionsForBounds(new Range<>(minToken, minToken));
Review Comment:
Because I'm not sure we can't move an sstable beyond its last key, I changed the caller to make sure we can work with that and changed the doc to state that possibility.
##########
src/java/org/apache/cassandra/io/sstable/format/SSTableReader.java:
##########
@@ -728,27 +728,104 @@ public List<PartitionPositionBounds> getPositionsForRanges(Collection<Range<Toke
for (Range<Token> range : Range.normalize(ranges))
{
assert !range.isWrapAround() || range.right.isMinimum();
- // truncate the range so it at most covers the sstable
AbstractBounds<PartitionPosition> bounds = Range.makeRowRange(range);
- PartitionPosition leftBound = bounds.left.compareTo(first) > 0 ? bounds.left : first.getToken().minKeyBound();
- PartitionPosition rightBound = bounds.right.isMinimum() ? last.getToken().maxKeyBound() : bounds.right;
+ PartitionPositionBounds pb = getPositionsForBounds(bounds);
+ if (pb != null)
+ positions.add(pb);
+ }
+ return positions;
+ }
- if (leftBound.compareTo(last) > 0 || rightBound.compareTo(first) < 0)
- continue;
+ /**
+ * Get a list of data positions in this SSTable that correspond to the given list of bounds. This method will remove
+ * non-covered intervals, but will not correct order or overlap in the supplied list, e.g. if bounds overlap, the
+ * result will be sections of the data file that repeat the same positions.
+ *
+ * @return A sorted list of [offset,end) pairs corresponding to the given boundsList in the datafile for this
+ * SSTable.
+ */
+ public List<PartitionPositionBounds> getPositionsForBoundsIterator(Iterator<AbstractBounds<PartitionPosition>> boundsList)
+ {
+ // use the index to determine a minimal section for each range
+ List<PartitionPositionBounds> positions = new ArrayList<>();
+ while (boundsList.hasNext())
+ {
+ AbstractBounds<PartitionPosition> bounds = boundsList.next();
+ PartitionPositionBounds pb = getPositionsForBounds(bounds);
+ if (pb != null)
+ positions.add(pb);
+ }
+ return positions;
+ }
- long left = getPosition(leftBound, Operator.GT);
- long right = (rightBound.compareTo(last) > 0)
- ? uncompressedLength()
- : getPosition(rightBound, Operator.GT);
+ /**
+ * Determine the data positions in this SSTable that cover the given bounds.
+ *
+ * @return An [offset,end) pair that cover the given bounds in the datafile for this SSTable.
+ */
+ public PartitionPositionBounds getPositionsForBounds(AbstractBounds<PartitionPosition> bounds)
+ {
+ long left = getPosition(bounds.left, bounds.inclusiveLeft() ? Operator.GE : Operator.GT);
+ // Note: getPosition will apply a moved start if the sstable is in MOVED_START state.
+ if (left < 0) // empty range
+ return null;
- if (left == right)
- // empty range
- continue;
+ long right = bounds.right.isMinimum() ? -1
+ : getPosition(bounds.right, bounds.inclusiveRight() ? Operator.GT
+ : Operator.GE);
+ if (right < 0) // right is beyond end
+ right = uncompressedLength(); // this should also be correct for EARLY readers
- assert left < right : String.format("Range=%s openReason=%s first=%s last=%s left=%d right=%d", range, openReason, first, last, left, right);
- positions.add(new PartitionPositionBounds(left, right));
+ if (left >= right)
+ {
+ // empty range
+ return null;
+ }
+
+ return new PartitionPositionBounds(left, right);
+ }
+
+ /**
+ * Return a (offset, end] pair that covers the whole file.
+ */
+ public PartitionPositionBounds getPositionsForFullRange()
+ {
+ if (openReason != OpenReason.MOVED_START)
+ return new PartitionPositionBounds(0, uncompressedLength());
+ else
+ {
+ // query a full range, so that the required adjustments can be applied
+ PartitionPosition minToken = getPartitioner().getMinimumToken().minKeyBound();
+ return getPositionsForBounds(new Range<>(minToken, minToken));
}
- return positions;
+ }
+
+ /**
+ * Calculate a total on-disk (compressed) size for the given partition positions.
Review Comment:
Added
##########
src/java/org/apache/cassandra/io/sstable/format/SSTableSimpleScanner.java:
##########
@@ -0,0 +1,185 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.cassandra.io.sstable.format;
+
+import java.util.Collection;
+import java.util.Iterator;
+import java.util.NoSuchElementException;
+import java.util.Set;
+import java.util.concurrent.atomic.AtomicBoolean;
+
+import com.google.common.collect.ImmutableSet;
+
+import org.apache.cassandra.db.DecoratedKey;
+import org.apache.cassandra.db.rows.UnfilteredRowIterator;
+import org.apache.cassandra.io.sstable.CorruptSSTableException;
+import org.apache.cassandra.io.sstable.ISSTableScanner;
+import org.apache.cassandra.io.sstable.SSTableIdentityIterator;
+import org.apache.cassandra.io.util.RandomAccessReader;
+import org.apache.cassandra.schema.TableMetadata;
+
+import static org.apache.cassandra.io.sstable.format.SSTableReader.PartitionPositionBounds;
+
+/// Simple SSTable scanner that reads sequentially through an SSTable without using the index.
+///
+/// This is a significant improvement for the performance of compaction over using the full-blown DataRange-capable
+/// [SSTableScanner] and enables correct calculation of data sizes to process.
+public class SSTableSimpleScanner
+implements ISSTableScanner
+{
+ private final AtomicBoolean isClosed = new AtomicBoolean(false);
+ private final RandomAccessReader dfile;
+ private final SSTableReader sstable;
+
+ private final Iterator<PartitionPositionBounds> rangeIterator;
+
+ private long bytesScannedInPreviousRanges;
+
+ private final long sizeInBytes;
+ private final long compressedSizeInBytes;
+
+ private long currentEndPosition;
+ private long currentStartPosition;
+
+ private SSTableIdentityIterator currentIterator;
+ private DecoratedKey lastKey;
+
+ public SSTableSimpleScanner(SSTableReader sstable,
Review Comment:
Added.
##########
src/java/org/apache/cassandra/io/sstable/format/SSTableSimpleScanner.java:
##########
@@ -0,0 +1,185 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.cassandra.io.sstable.format;
+
+import java.util.Collection;
+import java.util.Iterator;
+import java.util.NoSuchElementException;
+import java.util.Set;
+import java.util.concurrent.atomic.AtomicBoolean;
+
+import com.google.common.collect.ImmutableSet;
+
+import org.apache.cassandra.db.DecoratedKey;
+import org.apache.cassandra.db.rows.UnfilteredRowIterator;
+import org.apache.cassandra.io.sstable.CorruptSSTableException;
+import org.apache.cassandra.io.sstable.ISSTableScanner;
+import org.apache.cassandra.io.sstable.SSTableIdentityIterator;
+import org.apache.cassandra.io.util.RandomAccessReader;
+import org.apache.cassandra.schema.TableMetadata;
+
+import static org.apache.cassandra.io.sstable.format.SSTableReader.PartitionPositionBounds;
+
+/// Simple SSTable scanner that reads sequentially through an SSTable without using the index.
Review Comment:
Sent a [DISCUSS] message to the dev mailing list to ask about this.
##########
src/java/org/apache/cassandra/io/sstable/format/bti/BtiTableReader.java:
##########
@@ -382,27 +380,6 @@ public UnfilteredRowIterator rowIterator(FileDataInput dataFileInput,
return new SSTableIterator(this, dataFileInput, key, indexEntry, slices, selectedColumns, rowIndexFile);
}
- @Override
Review Comment:
Removed the methods in two commits: removing the factory methods and the necessary `makeBounds` helpers in one, and removing the ability to iterate multiple ranges in the second. If you prefer to leave the latter in place, let me know to revert the second commit.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: pr-unsubscribe@cassandra.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: pr-unsubscribe@cassandra.apache.org
For additional commands, e-mail: pr-help@cassandra.apache.org
Re: [PR] CASSANDRA-20092: Introduce SSTableSimpleScanner for compaction [cassandra]
Posted by "blambov (via GitHub)" <gi...@apache.org>.
blambov commented on code in PR #3695:
URL: https://github.com/apache/cassandra/pull/3695#discussion_r1867599495
##########
src/java/org/apache/cassandra/io/sstable/format/bti/BtiTableReader.java:
##########
@@ -382,27 +380,6 @@ public UnfilteredRowIterator rowIterator(FileDataInput dataFileInput,
return new SSTableIterator(this, dataFileInput, key, indexEntry, slices, selectedColumns, rowIndexFile);
}
- @Override
Review Comment:
Well, this caused a lot of failures (apparently wrap-around ranges are actually used with `DataRange`), so I reverted most of the changes and only removed the unused factory methods in the latest commit.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: pr-unsubscribe@cassandra.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: pr-unsubscribe@cassandra.apache.org
For additional commands, e-mail: pr-help@cassandra.apache.org
Re: [PR] CASSANDRA-20092: Introduce SSTableSimpleScanner for compaction [cassandra]
Posted by "smiklosovic (via GitHub)" <gi...@apache.org>.
smiklosovic closed pull request #3695: CASSANDRA-20092: Introduce SSTableSimpleScanner for compaction
URL: https://github.com/apache/cassandra/pull/3695
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: pr-unsubscribe@cassandra.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: pr-unsubscribe@cassandra.apache.org
For additional commands, e-mail: pr-help@cassandra.apache.org