Apache
/
hadoop
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

HDFS-16082. Atomic operations on exceptionsSinceLastBalance and failedTimesSinceLastSuccessfulBalance in Balancer (#3127)

This commit is contained in:
Viraj Jasani 2021-06-23 08:51:34 +05:30 committed by GitHub
parent 10b79a26fe
commit d9fbb3c508
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 13 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/balancer/Balancer.java
View File

@ -37,6 +37,7 @@ import java.util.List;
import java.util.Map; import java.util.Map;
import java.util.Set; import java.util.Set;
import java.util.concurrent.TimeUnit; import java.util.concurrent.TimeUnit;
import java.util.concurrent.atomic.AtomicInteger;
import org.apache.commons.lang3.builder.ToStringBuilder; import org.apache.commons.lang3.builder.ToStringBuilder;
import org.apache.hadoop.thirdparty.com.google.common.annotations.VisibleForTesting; import org.apache.hadoop.thirdparty.com.google.common.annotations.VisibleForTesting;
@ -211,8 +212,10 @@ public class Balancer {
@VisibleForTesting @VisibleForTesting
private static volatile boolean serviceRunning = false; private static volatile boolean serviceRunning = false;
private static volatile int exceptionsSinceLastBalance = 0; private static final AtomicInteger EXCEPTIONS_SINCE_LAST_BALANCE =
private static volatile int failedTimesSinceLastSuccessfulBalance = 0; new AtomicInteger(0);
private static final AtomicInteger
FAILED_TIMES_SINCE_LAST_SUCCESSFUL_BALANCE = new AtomicInteger(0);
private final Dispatcher dispatcher; private final Dispatcher dispatcher;
private final NameNodeConnector nnc; private final NameNodeConnector nnc;
@ -274,11 +277,11 @@ public class Balancer {
} }
static int getExceptionsSinceLastBalance() { static int getExceptionsSinceLastBalance() {
return exceptionsSinceLastBalance; return EXCEPTIONS_SINCE_LAST_BALANCE.get();
} }
static int getFailedTimesSinceLastSuccessfulBalance() { static int getFailedTimesSinceLastSuccessfulBalance() {
return failedTimesSinceLastSuccessfulBalance; return FAILED_TIMES_SINCE_LAST_SUCCESSFUL_BALANCE.get();
} }
/** /**
@ -866,20 +869,21 @@ public class Balancer {
int retCode = doBalance(namenodes, nsIds, p, conf); int retCode = doBalance(namenodes, nsIds, p, conf);
if (retCode < 0) { if (retCode < 0) {
LOG.info("Balance failed, error code: " + retCode); LOG.info("Balance failed, error code: " + retCode);
failedTimesSinceLastSuccessfulBalance++; FAILED_TIMES_SINCE_LAST_SUCCESSFUL_BALANCE.incrementAndGet();
} else { } else {
LOG.info("Balance succeed!"); LOG.info("Balance succeed!");
failedTimesSinceLastSuccessfulBalance = 0; FAILED_TIMES_SINCE_LAST_SUCCESSFUL_BALANCE.set(0);
} }
exceptionsSinceLastBalance = 0; EXCEPTIONS_SINCE_LAST_BALANCE.set(0);
} catch (Exception e) { } catch (Exception e) {
if (++exceptionsSinceLastBalance > retryOnException) { if (EXCEPTIONS_SINCE_LAST_BALANCE.incrementAndGet()
> retryOnException) {
// The caller will process and log the exception // The caller will process and log the exception
throw e; throw e;
} }
LOG.warn( LOG.warn(
"Encounter exception while do balance work. Already tried {} times", "Encounter exception while do balance work. Already tried {} times",
exceptionsSinceLastBalance, e); EXCEPTIONS_SINCE_LAST_BALANCE, e);
} }
// sleep for next round, will retry for next round when it's interrupted // sleep for next round, will retry for next round when it's interrupted