HDDS-1284. Adjust default values of pipline recovery for more resilient service restart. Contributed by Elek, Marton.

Author: elek

hadoop-hdds/common/src/main/java/org/apache/hadoop/hdds/scm/ScmConfigKeys.java

@@ -260,7 +260,7 @@ public final class ScmConfigKeys {
   public static final String OZONE_SCM_STALENODE_INTERVAL =
       "ozone.scm.stale.node.interval";
   public static final String OZONE_SCM_STALENODE_INTERVAL_DEFAULT =
-      "90s";
+      "5m";
 
   public static final String OZONE_SCM_HEARTBEAT_RPC_TIMEOUT =
       "ozone.scm.heartbeat.rpc-timeout";
@@ -331,7 +331,7 @@ public final class ScmConfigKeys {
       "ozone.scm.pipeline.destroy.timeout";
 
   public static final String OZONE_SCM_PIPELINE_DESTROY_TIMEOUT_DEFAULT =
-      "300s";
+      "66s";
 
   public static final String OZONE_SCM_PIPELINE_CREATION_INTERVAL =
       "ozone.scm.pipeline.creation.interval";

hadoop-hdds/common/src/main/resources/ozone-default.xml

@@ -1025,7 +1025,7 @@
   </property>
   <property>
     <name>ozone.scm.stale.node.interval</name>
-    <value>90s</value>
+    <value>5m</value>
     <tag>OZONE, MANAGEMENT</tag>
     <description>
       The interval for stale node flagging. Please
@@ -1264,7 +1264,7 @@
   </property>
   <property>
     <name>ozone.scm.pipeline.destroy.timeout</name>
-    <value>300s</value>
+    <value>66s</value>
     <tag>OZONE, SCM, PIPELINE</tag>
     <description>
       Once a pipeline is closed, SCM should wait for the above configured time