// Warning: scrubTimestamp MUST be the start timestamp of the hard delete transaction that
// triggers
// the scrubbing; we need this start timestamp to check whether the hard delete transaction was
// actually committed before we do any scrubbing
/* package */ void queueCellsForScrubbing(
Multimap<String, Cell> tableNameToCell, long scrubTimestamp) {
if (tableNameToCell.isEmpty()) {
return;
}
scrubberStore.queueCellsForScrubbing(tableNameToCell, scrubTimestamp, batchSizeSupplier.get());
}
/** @return number of cells read from _scrub table */
private int scrubSomeCells(
final TransactionManager txManager, int maxCellsToScrub, long maxScrubTimestamp) {
SortedMap<Long, Multimap<String, Cell>> scrubTimestampToTableNameToCell =
scrubberStore.getCellsToScrub(maxCellsToScrub, maxScrubTimestamp);
// Don't call expensive toString() if trace logging is off
if (log.isTraceEnabled()) {
log.trace("Attempting to scrub cells: " + scrubTimestampToTableNameToCell);
}
if (log.isInfoEnabled()) {
Set<String> tables = Sets.newHashSet();
for (Multimap<String, Cell> v : scrubTimestampToTableNameToCell.values()) {
tables.addAll(v.keySet());
}
log.info(
"Attempting to scrub "
+ scrubTimestampToTableNameToCell.size()
+ " cells from tables "
+ tables);
}
if (scrubTimestampToTableNameToCell.size() == 0) {
return 0; // No cells left to scrub
}
Multimap<Long, Cell> toRemoveFromScrubQueue = HashMultimap.create();
int numCellsReadFromScrubTable = 0;
for (Map.Entry<Long, Multimap<String, Cell>> entry :
scrubTimestampToTableNameToCell.entrySet()) {
final long scrubTimestamp = entry.getKey();
final Multimap<String, Cell> tableNameToCell = entry.getValue();
numCellsReadFromScrubTable += tableNameToCell.size();
long commitTimestamp = getCommitTimestampRollBackIfNecessary(scrubTimestamp, tableNameToCell);
if (commitTimestamp >= maxScrubTimestamp) {
// We cannot scrub this yet because not all transactions can read this value.
continue;
} else if (commitTimestamp != TransactionConstants.FAILED_COMMIT_TS) {
// This is CRITICAL; don't scrub if the hard delete transaction didn't actually finish
// (we still remove it from the _scrub table with the call to markCellsAsScrubbed though),
// or else we could cause permanent data loss if the hard delete transaction failed after
// queuing cells to scrub but before successfully committing
List<Future<Void>> scrubFutures = Lists.newArrayList();
for (final List<Entry<String, Cell>> batch :
Iterables.partition(tableNameToCell.entries(), batchSizeSupplier.get())) {
final Multimap<String, Cell> batchMultimap = HashMultimap.create();
for (Entry<String, Cell> e : batch) {
batchMultimap.put(e.getKey(), e.getValue());
}
scrubFutures.add(
exec.submit(
new Callable<Void>() {
@Override
public Void call() throws Exception {
scrubCells(
txManager,
batchMultimap,
scrubTimestamp,
aggressiveScrub
? TransactionType.AGGRESSIVE_HARD_DELETE
: TransactionType.HARD_DELETE);
return null;
}
}));
}
for (Future<Void> future : scrubFutures) {
Futures.getUnchecked(future);
}
}
toRemoveFromScrubQueue.putAll(scrubTimestamp, tableNameToCell.values());
}
Multimap<Cell, Long> cellToScrubTimestamp = HashMultimap.create();
scrubberStore.markCellsAsScrubbed(
Multimaps.invertFrom(toRemoveFromScrubQueue, cellToScrubTimestamp),
batchSizeSupplier.get());
if (log.isTraceEnabled()) {
log.trace("Finished scrubbing cells: " + scrubTimestampToTableNameToCell);
}
if (log.isInfoEnabled()) {
Set<String> tables = Sets.newHashSet();
for (Multimap<String, Cell> v : scrubTimestampToTableNameToCell.values()) {
tables.addAll(v.keySet());
}
log.info(
"Finished scrubbing "
+ scrubTimestampToTableNameToCell.size()
+ " cells from tables "
+ tables);
}
return numCellsReadFromScrubTable;
}