You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hbase.apache.org by zh...@apache.org on 2022/09/01 01:23:51 UTC
[hbase] branch master updated: HBASE-26809: Report client backoff time for server overloaded (#4729)
This is an automated email from the ASF dual-hosted git repository.
zhangduo pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/hbase.git
The following commit(s) were added to refs/heads/master by this push:
new 308cd729d23 HBASE-26809: Report client backoff time for server overloaded (#4729)
308cd729d23 is described below
commit 308cd729d23329e6d8d4b9c17a645180374b5962
Author: Bri Augenreich <bb...@vt.edu>
AuthorDate: Wed Aug 31 21:23:40 2022 -0400
HBASE-26809: Report client backoff time for server overloaded (#4729)
Co-authored-by: Briana Augenreich <ba...@hubspot.com>
Signed-off-by: Duo Zhang <zh...@apache.org>
---
.../apache/hadoop/hbase/client/AsyncBatchRpcRetryingCaller.java | 5 +++++
.../org/apache/hadoop/hbase/client/AsyncRpcRetryingCaller.java | 4 ++++
.../hbase/client/AsyncScanSingleRegionRpcRetryingCaller.java | 8 ++++++++
.../java/org/apache/hadoop/hbase/client/MetricsConnection.java | 8 ++++++++
4 files changed, 25 insertions(+)
diff --git a/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncBatchRpcRetryingCaller.java b/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncBatchRpcRetryingCaller.java
index 0798915c08d..49cf7589207 100644
--- a/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncBatchRpcRetryingCaller.java
+++ b/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncBatchRpcRetryingCaller.java
@@ -487,6 +487,11 @@ class AsyncBatchRpcRetryingCaller<T> {
} else {
delayNs = getPauseTime(pauseNsToUse, tries - 1);
}
+
+ if (isServerOverloaded) {
+ Optional<MetricsConnection> metrics = conn.getConnectionMetrics();
+ metrics.ifPresent(m -> m.incrementServerOverloadedBackoffTime(delayNs, TimeUnit.NANOSECONDS));
+ }
retryTimer.newTimeout(t -> groupAndSend(actions, tries + 1), delayNs, TimeUnit.NANOSECONDS);
}
diff --git a/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncRpcRetryingCaller.java b/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncRpcRetryingCaller.java
index a19d3b039f1..04e22710838 100644
--- a/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncRpcRetryingCaller.java
+++ b/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncRpcRetryingCaller.java
@@ -139,6 +139,10 @@ public abstract class AsyncRpcRetryingCaller<T> {
delayNs = getPauseTime(pauseNsToUse, tries - 1);
}
tries++;
+ if (HBaseServerException.isServerOverloaded(error)) {
+ Optional<MetricsConnection> metrics = conn.getConnectionMetrics();
+ metrics.ifPresent(m -> m.incrementServerOverloadedBackoffTime(delayNs, TimeUnit.NANOSECONDS));
+ }
retryTimer.newTimeout(t -> doCall(), delayNs, TimeUnit.NANOSECONDS);
}
diff --git a/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncScanSingleRegionRpcRetryingCaller.java b/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncScanSingleRegionRpcRetryingCaller.java
index dbaae5c26e2..3ef7b9b6ccc 100644
--- a/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncScanSingleRegionRpcRetryingCaller.java
+++ b/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncScanSingleRegionRpcRetryingCaller.java
@@ -113,6 +113,8 @@ class AsyncScanSingleRegionRpcRetryingCaller {
private final Runnable completeWhenNoMoreResultsInRegion;
+ private final AsyncConnectionImpl conn;
+
private final CompletableFuture<Boolean> future;
private final HBaseRpcController controller;
@@ -318,6 +320,7 @@ class AsyncScanSingleRegionRpcRetryingCaller {
long pauseNsForServerOverloaded, int maxAttempts, long scanTimeoutNs, long rpcTimeoutNs,
int startLogErrorsCnt) {
this.retryTimer = retryTimer;
+ this.conn = conn;
this.scan = scan;
this.scanMetrics = scanMetrics;
this.scannerId = scannerId;
@@ -441,6 +444,11 @@ class AsyncScanSingleRegionRpcRetryingCaller {
return;
}
tries++;
+
+ if (HBaseServerException.isServerOverloaded(error)) {
+ Optional<MetricsConnection> metrics = conn.getConnectionMetrics();
+ metrics.ifPresent(m -> m.incrementServerOverloadedBackoffTime(delayNs, TimeUnit.NANOSECONDS));
+ }
retryTimer.newTimeout(t -> call(), delayNs, TimeUnit.NANOSECONDS);
}
diff --git a/hbase-client/src/main/java/org/apache/hadoop/hbase/client/MetricsConnection.java b/hbase-client/src/main/java/org/apache/hadoop/hbase/client/MetricsConnection.java
index dc452bcd9d9..f844c47e406 100644
--- a/hbase-client/src/main/java/org/apache/hadoop/hbase/client/MetricsConnection.java
+++ b/hbase-client/src/main/java/org/apache/hadoop/hbase/client/MetricsConnection.java
@@ -315,6 +315,7 @@ public class MetricsConnection implements StatisticTrackable {
protected final Histogram numActionsPerServerHist;
protected final Counter nsLookups;
protected final Counter nsLookupsFailed;
+ protected final Timer overloadedBackoffTimer;
// dynamic metrics
@@ -376,6 +377,8 @@ public class MetricsConnection implements StatisticTrackable {
registry.histogram(name(MetricsConnection.class, "numActionsPerServer", scope));
this.nsLookups = registry.counter(name(this.getClass(), NS_LOOKUPS, scope));
this.nsLookupsFailed = registry.counter(name(this.getClass(), NS_LOOKUPS_FAILED, scope));
+ this.overloadedBackoffTimer =
+ registry.timer(name(this.getClass(), "overloadedBackoffDurationMs", scope));
this.reporter = JmxReporter.forRegistry(this.registry).build();
this.reporter.start();
@@ -449,6 +452,11 @@ public class MetricsConnection implements StatisticTrackable {
this.runnerStats.updateDelayInterval(interval);
}
+ /** Update the overloaded backoff time **/
+ public void incrementServerOverloadedBackoffTime(long time, TimeUnit timeUnit) {
+ overloadedBackoffTimer.update(time, timeUnit);
+ }
+
/**
* Get a metric for {@code key} from {@code map}, or create it with {@code factory}.
*/