Skip to content

should not waitBackOff when replicating open fragment #4593

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

Closed
SongOf wants to merge 1 commit into from
Closed

should not waitBackOff when replicating open fragment #4593

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
34 changes: 21 additions & 13 deletions bookkeeper-server/src/main/java/org/apache/bookkeeper/replication/ReplicationWorker.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -111,6 +111,11 @@ public class ReplicationWorker implements Runnable {
private final long baseBackoffForLockReleaseOfFailedLedger;
private final BiConsumer<Long, Long> onReadEntryFailureCallback;
private final LedgerManager ledgerManager;
enum ReplicateResult {
SUCCESS,
FAILED,
SKIP_OPEN_FRAGMENT
}

// Expose Stats
private final StatsLogger statsLogger;
Expand Down Expand Up @@ -246,7 +251,7 @@ public void run() {
workerRunning = true;
while (workerRunning) {
try {
if (!rereplicate()) {
if (rereplicate() == ReplicateResult.FAILED) {
LOG.warn("failed while replicating fragments");
waitBackOffTime(rwRereplicateBackoffMs);
}
Expand Down Expand Up @@ -290,24 +295,27 @@ private static void waitBackOffTime(long backoffMs) {
* Replicates the under replicated fragments from failed bookie ledger to
* targetBookie.
*/
private boolean rereplicate() throws InterruptedException, BKException,
private ReplicateResult rereplicate() throws InterruptedException, BKException,
UnavailableException {
long ledgerIdToReplicate = underreplicationManager
.getLedgerToRereplicate();

Stopwatch stopwatch = Stopwatch.createStarted();
boolean success = false;

ReplicateResult result = ReplicateResult.FAILED;
try {
success = rereplicate(ledgerIdToReplicate);
result = rereplicate(ledgerIdToReplicate);
} finally {
long latencyMillis = stopwatch.stop().elapsed(TimeUnit.MILLISECONDS);
if (success) {
if (result == ReplicateResult.SUCCESS) {
rereplicateOpStats.registerSuccessfulEvent(latencyMillis, TimeUnit.MILLISECONDS);
} else {
} else if (result == ReplicateResult.FAILED) {
rereplicateOpStats.registerFailedEvent(latencyMillis, TimeUnit.MILLISECONDS);
} else {
LOG.info("Replication of ledger with open fragment is skipped, ledgerId is {}.", ledgerIdToReplicate);
}
}
return success;
return result;
}

private void logBKExceptionAndReleaseLedger(BKException e, long ledgerIdToReplicate)
Expand Down Expand Up @@ -440,7 +448,7 @@ private Set<LedgerFragment> getNeedRepairedPlacementNotAdheringFragments(LedgerH
}

@SuppressFBWarnings("RCN_REDUNDANT_NULLCHECK_WOULD_HAVE_BEEN_A_NPE")
private boolean rereplicate(long ledgerIdToReplicate) throws InterruptedException, BKException,
private ReplicateResult rereplicate(long ledgerIdToReplicate) throws InterruptedException, BKException,
UnavailableException {
if (LOG.isDebugEnabled()) {
LOG.debug("Going to replicate the fragments of the ledger: {}", ledgerIdToReplicate);
Expand Down Expand Up @@ -495,21 +503,21 @@ private boolean rereplicate(long ledgerIdToReplicate) throws InterruptedExceptio
if (foundOpenFragments || isLastSegmentOpenAndMissingBookies(lh)) {
deferLedgerLockRelease = true;
deferLedgerLockRelease(ledgerIdToReplicate);
return false;
return ReplicateResult.SKIP_OPEN_FRAGMENT;
}

fragments = getUnderreplicatedFragments(lh, conf.getAuditorLedgerVerificationPercentage());
if (fragments.size() == 0) {
LOG.info("Ledger replicated successfully. ledger id is: " + ledgerIdToReplicate);
underreplicationManager.markLedgerReplicated(ledgerIdToReplicate);
return true;
return ReplicateResult.SUCCESS;
} else {
deferLedgerLockRelease = true;
deferLedgerLockReleaseOfFailedLedger(ledgerIdToReplicate);
numDeferLedgerLockReleaseOfFailedLedger.inc();
// Releasing the underReplication ledger lock and compete
// for the replication again for the pending fragments
return false;
return ReplicateResult.FAILED;
}

} catch (BKNoSuchLedgerExistsOnMetadataServerException e) {
Expand All @@ -520,13 +528,13 @@ private boolean rereplicate(long ledgerIdToReplicate) throws InterruptedExceptio
+ "So, no harm to continue", ledgerIdToReplicate);
underreplicationManager.markLedgerReplicated(ledgerIdToReplicate);
getExceptionCounter("BKNoSuchLedgerExistsOnMetadataServerException").inc();
return false;
return ReplicateResult.FAILED;
} catch (BKNotEnoughBookiesException e) {
logBKExceptionAndReleaseLedger(e, ledgerIdToReplicate);
throw e;
} catch (BKException e) {
logBKExceptionAndReleaseLedger(e, ledgerIdToReplicate);
return false;
return ReplicateResult.FAILED;
} finally {
// we make sure we always release the underreplicated lock, unless we decided to defer it. If the lock has
// already been released, this is a no-op
Expand Down
42 changes: 40 additions & 2 deletions bookkeeper-server/src/test/java/org/apache/bookkeeper/replication/TestReplicationWorker.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1248,6 +1248,44 @@ public void testReplicateEmptyOpenStateLedger() throws Exception {
}
}

@Test
public void testReplicateOpenStateLedger() throws Exception {
LedgerHandle lh = bkc.createLedger(3, 3, BookKeeper.DigestType.CRC32,
TESTPASSWD);
for (int i = 0; i < 10; i++) {
lh.addEntry(data);
}

BookieId replicaToKill = lh.getLedgerMetadata().getAllEnsembles().get(0L).get(0);
LOG.info("Killing Bookie : {}", replicaToKill);
killBookie(replicaToKill);

ReplicationWorker rw = new ReplicationWorker(baseConf);
try {
underReplicationManager.markLedgerUnderreplicated(lh.getId(),
replicaToKill.toString());
while (!ReplicationTestUtil.isLedgerInUnderReplication(zkc, lh
.getId(), basePath)) {
Thread.sleep(100);
}

final Method rereplicate = rw.getClass().getDeclaredMethod("rereplicate", long.class);
rereplicate.setAccessible(true);
assertFalse(lh.getLedgerMetadata().isClosed());
Object result = rereplicate.invoke(rw, lh.getId());
assertEquals(result, ReplicationWorker.ReplicateResult.SKIP_OPEN_FRAGMENT);
lh.close();
assertTrue(lh.getLedgerMetadata().isClosed());
result = rereplicate.invoke(rw, lh.getId());
assertEquals(result, ReplicationWorker.ReplicateResult.FAILED);
BookieId newBkAddr = startNewBookieAndReturnBookieId();
LOG.info("New Bookie addr : {}", newBkAddr);
result = rereplicate.invoke(rw, lh.getId());
assertEquals(result, ReplicationWorker.ReplicateResult.SUCCESS);
} finally {
rw.shutdown();
}
}
@Test
public void testRepairedNotAdheringPlacementPolicyLedgerFragmentsOnRack() throws Exception {
testRepairedNotAdheringPlacementPolicyLedgerFragments(RackawareEnsemblePlacementPolicy.class, null);
Expand Down Expand Up @@ -1275,14 +1313,14 @@ public void testReplicationStats() throws Exception {
1, numDeferLedgerLockReleaseOfFailedLedgerCounter.get().longValue());

if (first) {
assertFalse((boolean) result);
assertEquals(result, ReplicationWorker.ReplicateResult.FAILED);
assertEquals("NUM_FULL_OR_PARTIAL_LEDGERS_REPLICATED",
0, numLedgersReplicatedCounter.get().longValue());
assertEquals("NUM_NOT_ADHERING_PLACEMENT_LEDGERS_REPLICATED",
0, numNotAdheringPlacementLedgersCounter.get().longValue());

} else {
assertTrue((boolean) result);
assertEquals(result, ReplicationWorker.ReplicateResult.SUCCESS);
assertEquals("NUM_FULL_OR_PARTIAL_LEDGERS_REPLICATED",
1, numLedgersReplicatedCounter.get().longValue());
assertEquals("NUM_NOT_ADHERING_PLACEMENT_LEDGERS_REPLICATED",
Expand Down