[
https://issues.apache.org/jira/browse/HDFS-16671?focusedWorklogId=794399&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-794399
]
ASF GitHub Bot logged work on HDFS-16671:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 22/Jul/22 22:17
Start Date: 22/Jul/22 22:17
Worklog Time Spent: 10m
Work Description: ayushtkn commented on code in PR #4597:
URL: https://github.com/apache/hadoop/pull/4597#discussion_r928012699
##########
hadoop-hdfs-project/hadoop-hdfs-rbf/src/main/java/org/apache/hadoop/hdfs/server/federation/router/RBFConfigKeys.java:
##########
@@ -354,6 +354,10 @@ public class RBFConfigKeys extends
CommonConfigurationKeysPublic {
NoRouterRpcFairnessPolicyController.class;
public static final String DFS_ROUTER_FAIR_HANDLER_COUNT_KEY_PREFIX =
FEDERATION_ROUTER_FAIRNESS_PREFIX + "handler.count.";
+ public static final String DFS_ROUTER_FAIRNESS_ACQUIRE_TIMEOUT =
+ FEDERATION_ROUTER_FAIRNESS_PREFIX + "acquire.timeout";
+ public static final long DFS_ROUTER_FAIRNESS_ACQUIRE_TIMEOUT_MS_DEFAULT =
Review Comment:
The variable name should be similar to the original config name +
``_DEFAULT``, remove the ``_MS_`` in middle.
Should be ``DFS_ROUTER_FAIRNESS_ACQUIRE_TIMEOUT_DEFAULT``
##########
hadoop-hdfs-project/hadoop-hdfs-rbf/src/main/java/org/apache/hadoop/hdfs/server/federation/fairness/AbstractRouterRpcFairnessPolicyController.java:
##########
@@ -42,15 +45,22 @@
/** Hash table to hold semaphore for each configured name service. */
private Map<String, Semaphore> permits;
+ private long acquireTimeoutMs =
DFS_ROUTER_FAIRNESS_ACQUIRE_TIMEOUT_MS_DEFAULT;
+
public void init(Configuration conf) {
this.permits = new HashMap<>();
+ long timeoutMs = conf.getTimeDuration(DFS_ROUTER_FAIRNESS_ACQUIRE_TIMEOUT,
+ DFS_ROUTER_FAIRNESS_ACQUIRE_TIMEOUT_MS_DEFAULT, TimeUnit.MILLISECONDS);
+ if (timeoutMs >= 0) {
+ acquireTimeoutMs = timeoutMs;
Review Comment:
if there is an invalid entry configured and we are moving to using the
default value. We should atleast have a warn log. Kind of ``Invalid value -1
configured for dfs.... should be greater than or equal to 0, Using default
value of : 1s instead``
something like this
##########
hadoop-hdfs-project/hadoop-hdfs-rbf/src/test/java/org/apache/hadoop/hdfs/server/federation/fairness/TestRouterRpcFairnessPolicyController.java:
##########
@@ -83,6 +87,29 @@ public void testHandlerAllocationPreconfigured() {
assertFalse(routerRpcFairnessPolicyController.acquirePermit(CONCURRENT_NS));
}
+ @Test
+ public void testAcquireTimeout() {
+ Configuration conf = createConf(40);
+ conf.setInt(DFS_ROUTER_FAIR_HANDLER_COUNT_KEY_PREFIX + "ns1", 30);
+ conf.setTimeDuration(DFS_ROUTER_FAIRNESS_ACQUIRE_TIMEOUT, 100,
TimeUnit.MILLISECONDS);
+ RouterRpcFairnessPolicyController routerRpcFairnessPolicyController =
+ FederationUtil.newFairnessPolicyController(conf);
+
+ // ns1 should have 30 permits allocated
+ for (int i = 0; i < 30; i++) {
+ assertTrue(routerRpcFairnessPolicyController.acquirePermit("ns1"));
+ }
+ long acquireBeginTimeMs = Time.monotonicNow();
+ assertFalse(routerRpcFairnessPolicyController.acquirePermit("ns1"));
+ long acquireEndTimeMs = Time.monotonicNow();
+
+ long acquireTimeMs = acquireEndTimeMs - acquireBeginTimeMs;
+
+ // There are some other operations, so acquireTimeMs >= 100ms.
+ assertTrue(acquireTimeMs >= 100);
+ assertTrue(acquireTimeMs < 100 + 50);
Review Comment:
Is this 50 added as safe margin, kind of that other operations won't take
more than 50ms? In that case I doubt this test is gonna go flaky in future,
Double check to confirm it isn't gonna cross the safe limit of 50 ever.
Issue Time Tracking
-------------------
Worklog Id: (was: 794399)
Time Spent: 1.5h (was: 1h 20m)
> RBF: RouterRpcFairnessPolicyController supports configurable permit acquire
> timeout
> -----------------------------------------------------------------------------------
>
> Key: HDFS-16671
> URL: https://issues.apache.org/jira/browse/HDFS-16671
> Project: Hadoop HDFS
> Issue Type: Improvement
> Reporter: ZanderXu
> Assignee: ZanderXu
> Priority: Major
> Labels: pull-request-available
> Time Spent: 1.5h
> Remaining Estimate: 0h
>
> RouterRpcFairnessPolicyController supports configurable permit acquire
> timeout. Hardcode 1s is very long, and it has causes an incident in our prod
> environment when one nameserivce is busy.
> And the optimal timeout maybe should be less than p50(avgTime).
> And all handlers in RBF is waiting to acquire the permit of the busy ns.
> {code:java}
> "IPC Server handler 12 on default port 8888" #2370 daemon prio=5 os_prio=0
> tid=? nid=? waiting on condition [?]
> java.lang.Thread.State: TIMED_WAITING (parking)
> at sun.misc.Unsafe.park(Native Method)
> - parking to wait for <?> (a
> java.util.concurrent.Semaphore$NonfairSync)
> at
> java.util.concurrent.locks.LockSupport.parkNanos(LockSupport.java:215)
> at
> java.util.concurrent.locks.AbstractQueuedSynchronizer.doAcquireSharedNanos(AbstractQueuedSynchronizer.java:1037)
> at
> java.util.concurrent.locks.AbstractQueuedSynchronizer.tryAcquireSharedNanos(AbstractQueuedSynchronizer.java:1328)
> at java.util.concurrent.Semaphore.tryAcquire(Semaphore.java:409)
> at
> org.apache.hadoop.hdfs.server.federation.fairness.AbstractRouterRpcFairnessPolicyController.acquirePermit(AbstractRouterRpcFairnessPolicyController.java:56)
> at
> org.apache.hadoop.hdfs.server.federation.fairness.DynamicRouterRpcFairnessPolicyController.acquirePermit(DynamicRouterRpcFairnessPolicyController.java:123)
> at
> org.apache.hadoop.hdfs.server.federation.router.RouterRpcClient.acquirePermit(RouterRpcClient.java:1500)
> {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
