Package org.apache.cassandra.index.sai.disk

Interface PerColumnIndexWriter

All Known Implementing Classes:

MemtableIndexWriter, SSTableIndexWriter

public interface PerColumnIndexWriter

Creates the on-disk components for a given column index.

Method Summary

Modifier and Type	Method	Description
void	abort(Throwable cause)	Aborts accumulating data.
void	addRow(PrimaryKey key, Row row, long sstableRowId)	Adds a row to this index.
void	complete(com.google.common.base.Stopwatch stopwatch)	Builds on-disk index data structures from accumulated data, moves them all to the filesystem, and fsync created files.
void	onSSTableWriterSwitched(com.google.common.base.Stopwatch stopwatch)	Called when current SSTable writer is switched during sharded compaction to free any in-memory resources associated with the SSTable for current index without waiting for full transaction to complete

Method Details

addRow

void addRow(PrimaryKey key,
 Row row,
 long sstableRowId)
     throws IOException

Adds a row to this index.

Note: Callers are responsible for ensuring that rows are added in PrimaryKey order.

Throws:

IOException

complete

void complete(com.google.common.base.Stopwatch stopwatch)
       throws IOException

Builds on-disk index data structures from accumulated data, moves them all to the filesystem, and fsync created files.

Throws:

IOException

onSSTableWriterSwitched

void onSSTableWriterSwitched(com.google.common.base.Stopwatch stopwatch)
                      throws IOException

Called when current SSTable writer is switched during sharded compaction to free any in-memory resources associated with the SSTable for current index without waiting for full transaction to complete

Throws:

IOException

abort

void abort(Throwable cause)

Aborts accumulating data. Allows to clean up resources on error.

Note: Implementations should be idempotent, i.e. safe to call multiple times without producing undesirable side-effects.