Merge "Tune OpenDaylight parameters"

5 files changed, 283 insertions, 0 deletions

diff --git a/kubernetes/sdnc/resources/config/conf/akka.conf b/kubernetes/sdnc/resources/config/conf/akka.conf
new file mode 100644
index 0000000000..8440a9de56
--- /dev/null
+++ b/kubernetes/sdnc/resources/config/conf/akka.conf
@@ -0,0 +1,80 @@
+
+odl-cluster-data {
+  akka {
+    remote {
+      artery {
+        enabled = off
+        canonical.hostname = "127.0.0.1"
+        canonical.port = 2550
+      }
+      netty.tcp {
+        hostname = "127.0.0.1"
+        port = 2550
+      }
+
+      use-passive-connections = off
+      # when under load we might trip a false positive on the failure detector
+      # transport-failure-detector {
+      # heartbeat-interval = 4 s
+      # acceptable-heartbeat-pause = 16s
+      # }
+    }
+
+    actor {
+      debug {
+        autoreceive = on
+        lifecycle = on
+        unhandled = on
+        fsm = on
+        event-stream = on
+      }
+    }
+
+    cluster {
+      # Remove ".tcp" when using artery.
+      seed-nodes = ["akka.tcp://opendaylight-cluster-data@127.0.0.1:2550"]
+
+      seed-node-timeout = {{.Values.config.odl.akka.seedNodeTimeout}}
+
+      roles = ["member-1"]
+
+    }
+
+    persistence {
+      # By default the snapshots/journal directories live in KARAF_HOME. You can choose to put it somewhere else by
+      # modifying the following two properties. The directory location specified may be a relative or absolute path.
+      # The relative path is always relative to KARAF_HOME.
+
+      # snapshot-store.local.dir = "target/snapshots"
+      # journal.leveldb.dir = "target/journal"
+
+      journal {
+        leveldb {
+            # Set native = off to use a Java-only implementation of leveldb.
+            # Note that the Java-only version is not currently considered by Akka to be production quality.
+
+            # native = off
+        }
+
+        journal-plugin-fallback {
+          circuit-breaker {
+              max-failures = {{.Values.config.odl.akka.circuitBreaker.maxFailures}}
+              call-timeout = {{.Values.config.odl.akka.circuitBreaker.callTimeout}}
+              reset-timeout = {{.Values.config.odl.akka.circuitBreaker.resetTimeout}}
+          }
+          recovery-event-timeout = {{.Values.config.odl.akka.recoveryEventTimeout}}
+        }
+
+        snapshot-store-plugin-fallback {
+          circuit-breaker {
+            max-failures = {{.Values.config.odl.akka.circuitBreaker.maxFailures}}
+            call-timeout = {{.Values.config.odl.akka.circuitBreaker.callTimeout}}
+            reset-timeout = {{.Values.config.odl.akka.circuitBreaker.resetTimeout}}
+          }
+          recovery-event-timeout = {{.Values.config.odl.akka.recoveryEventTimeout}}
+        }
+      }
+    }
+  }
+}
+
diff --git a/kubernetes/sdnc/resources/config/conf/org.opendaylight.controller.cluster.datastore.cfg b/kubernetes/sdnc/resources/config/conf/org.opendaylight.controller.cluster.datastore.cfg
new file mode 100644
index 0000000000..29dd0e54dd
--- /dev/null
+++ b/kubernetes/sdnc/resources/config/conf/org.opendaylight.controller.cluster.datastore.cfg
@@ -0,0 +1,107 @@
+# This file specifies property settings for the clustered data store to control its behavior. A
+# property may be applied to every data store type ("config" and "operational") or can be customized
+# differently for each data store type by prefixing the data store type + '.'. For example, specifying
+# the "shard-election-timeout-factor" property would be applied to both data stores whereas specifying
+# "operational.shard-election-timeout-factor" would only apply to the "operational" data store. Similarly,
+# specifying "config.shard-election-timeout-factor" would only apply to the "config" data store.
+
+# The multiplication factor to be used to determine shard election timeout. The shard election timeout
+# is determined by multiplying shardHeartbeatIntervalInMillis with the shardElectionTimeoutFactor.
+shard-election-timeout-factor=20
+
+# The interval at which a shard will send a heart beat message to its remote shard.
+#shard-heartbeat-interval-in-millis=500
+
+# The amount by which to divide election timeout in case of a candidate. This serves as a counter-balance
+# to shard-election-timeout-factor. The default value is 1, i.e. election timeout is the same in all
+# situations.
+#shard-candidate-election-timeout-divisor=1
+
+# The maximum amount of time to wait for a shard to elect a leader before failing an operation (eg transaction create).
+#shard-leader-election-timeout-in-seconds=30
+
+# Enable or disable data persistence.
+#persistent=true
+
+# Disable persistence for the operational data store by default.
+operational.persistent=false
+
+# The maximum amount of time a shard transaction can be idle without receiving any messages before it self-destructs.
+#shard-transaction-idle-timeout-in-minutes=10
+
+# The maximum amount of time a shard transaction three-phase commit can be idle without receiving the
+# next messages before it aborts the transaction.
+#shard-transaction-commit-timeout-in-seconds=30
+
+# The maximum allowed capacity for each shard's transaction commit queue.
+#shard-transaction-commit-queue-capacity=20000
+
+# The maximum amount of time to wait for a shard to initialize from persistence on startup before
+# failing an operation (eg transaction create and change listener registration).
+#shard-initialization-timeout-in-seconds=300
+
+# The minimum number of entries to be present in the in-memory journal log before a snapshot is to be taken.
+#shard-snapshot-batch-count=20000
+
+# The percentage of Runtime.totalMemory() used by the in-memory journal log before a snapshot is to be taken.
+#shard-snapshot-data-threshold-percentage=12
+
+# The interval at which the leader of the shard will check if its majority followers are active and
+# term itself as isolated.
+#shard-isolated-leader-check-interval-in-millis=5000
+
+# The number of transaction modification operations (put, merge, delete) to batch before sending to the
+# shard transaction actor. Batching improves performance as less modifications messages are sent to the
+# actor and thus lessens the chance that the transaction actor's mailbox queue could get full.
+#shard-batched-modification-count=1000
+
+# The maximum amount of time for akka operations (remote or local) to complete before failing.
+#operation-timeout-in-seconds=5
+
+# The initial number of transactions per second that are allowed before the data store should begin
+# applying back pressure. This number is only used as an initial guidance, subsequently the datastore
+# measures the latency for a commit and auto-adjusts the rate limit.
+#transaction-creation-initial-rate-limit=100
+
+# The maximum thread pool size for each shard's data store data change notification executor.
+#max-shard-data-change-executor-pool-size=20
+
+# The maximum queue size for each shard's data store data change notification executor.
+#max-shard-data-change-executor-queue-size=1000
+
+# The maximum queue size for each shard's data store data change listener.
+#max-shard-data-change-listener-queue-size=1000
+
+# The maximum queue size for each shard's data store executor.
+#max-shard-data-store-executor-queue-size=5000
+
+# A fully qualified java class name. The class should implement
+# org.opendaylight.controller.cluster.raft.policy.RaftPolicy. This java class should be
+# accessible to the distributed data store OSGi module so that it can be dynamically loaded via
+# reflection. For now let's assume that these classes to customize raft behaviors should be
+# present in the distributed data store module itself. If this property is set to a class which
+# cannot be found then the default raft policy will be applied
+#custom-raft-policy-implementation=
+
+# When fragmenting messages thru the akka remoting framework, this is the maximum size in bytes
+# for a message slice.
+#maximum-message-slice-size=20480000
+
+# Enable tell-based protocol between frontend (applications) and backend (shards). Using this protocol
+# should avoid AskTimeoutExceptions seen under heavy load. Defaults to false (use ask-based protocol).
+#use-tell-based-protocol=true
+
+# Tune the maximum number of entries a follower is allowed to lag behind the leader before it is
+# considered out-of-sync. This flag may require tuning in face of a large number of small transactions.
+#sync-index-threshold=10
+
+# Record new transaction allocation stack trace, useful for debugging.  This makes the log include
+# the stack trace of the creator of the Tx when there is an exception when the transaction is submitted 
+# (e.g. for a failed validation).  Defaults to false due to performance impact.
+#transaction-debug-context-enabled=true
+persistent-actor-restart-min-backoff-in-seconds={{.Values.config.odl.datastore.persistentActorRestartMinBackoffInSeconds}}
+persistent-actor-restart-max-backoff-in-seconds={{.Values.config.odl.datastore.persistentActorRestartMaxBackoffInSeconds}}
+persistent-actor-restart-reset-backoff-in-seconds={{.Values.config.odl.datastore.persistentActorRestartResetBackoffInSeconds}}
+shard-transaction-commit-timeout-in-seconds={{.Values.config.odl.datastore.shardTransactionCommitTimeoutInSeconds}}
+shard-isolated-leader-check-interval-in-millis={{.Values.config.odl.datastore.shardIsolatedLeaderCheckIntervalInMillis}}
+operation-timeout-in-seconds={{.Values.config.odl.datastore.operationTimeoutInSeconds}}
diff --git a/kubernetes/sdnc/resources/config/conf/setenv b/kubernetes/sdnc/resources/config/conf/setenv
new file mode 100644
index 0000000000..7476e6849a
--- /dev/null
+++ b/kubernetes/sdnc/resources/config/conf/setenv
@@ -0,0 +1,61 @@
+#!/bin/sh
+#
+#    Licensed to the Apache Software Foundation (ASF) under one or more
+#    contributor license agreements.  See the NOTICE file distributed with
+#    this work for additional information regarding copyright ownership.
+#    The ASF licenses this file to You under the Apache License, Version 2.0
+#    (the "License"); you may not use this file except in compliance with
+#    the License.  You may obtain a copy of the License at
+#
+#       http://www.apache.org/licenses/LICENSE-2.0
+#
+#    Unless required by applicable law or agreed to in writing, software
+#    distributed under the License is distributed on an "AS IS" BASIS,
+#    WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#    See the License for the specific language governing permissions and
+#    limitations under the License.
+#
+
+#
+# handle specific scripts; the SCRIPT_NAME is exactly the name of the Karaf
+# script: client, instance, shell, start, status, stop, karaf
+#
+# if [ "${KARAF_SCRIPT}" == "SCRIPT_NAME" ]; then
+#   Actions go here...
+# fi
+
+#
+# general settings which should be applied for all scripts go here; please keep
+# in mind that it is possible that scripts might be executed more than once, e.g.
+# in example of the start script where the start script is executed first and the
+# karaf script afterwards.
+#
+
+#
+# The following section shows the possible configuration options for the default 
+# karaf scripts
+#
+# export JAVA_HOME # Location of Java installation
+# export JAVA_MIN_MEM # Minimum memory for the JVM
+# export JAVA_MAX_MEM # Maximum memory for the JVM
+# export JAVA_PERM_MEM # Minimum perm memory for the JVM
+# export JAVA_MAX_PERM_MEM # Maximum perm memory for the JVM
+# export EXTRA_JAVA_OPTS # Additional JVM options
+# export KARAF_HOME # Karaf home folder
+# export KARAF_DATA # Karaf data folder
+# export KARAF_BASE # Karaf base folder
+# export KARAF_ETC  # Karaf etc  folder
+# export KARAF_SYSTEM_OPTS # First citizen Karaf options
+# export KARAF_OPTS # Additional available Karaf options
+# export KARAF_DEBUG # Enable debug mode
+# export KARAF_REDIRECT # Enable/set the std/err redirection when using bin/start
+# export KARAF_NOROOT # Prevent execution as root if set to true
+if [ "x$JAVA_MAX_MEM" = "x" ]; then
+    export JAVA_MAX_MEM="2048m"
+fi
+
+EXTRA_JAVA_OPTS: "-XX:+UseG1GC -XX:MaxGCPauseMillis={{.Values.config.odl.javaOptions.maxGCPauseMillis}} \
+ -XX:ParallelGCThreads={{.Values.config.odl.javaOptions.parallelGCThreads}} -XX:+ParallelRefProcEnabled \
+ -XX:+UseStringDeduplication -XX:+PrintGC -XX:+PrintGCDateStamps -XX:+PrintGCDetails \
+ -XX:+PrintGCTimeStamps -XX:+UseGCLogFileRotation \
+ -XX:NumberOfGCLogFiles={{.Values.config.odl.javaOptions.numberGGLogFiles}} -Xloggc:/var/log/onap/sdnc/gc-%t.log"
diff --git a/kubernetes/sdnc/templates/statefulset.yaml b/kubernetes/sdnc/templates/statefulset.yaml
index d60319fa1c..e3ac50fc61 100644
--- a/kubernetes/sdnc/templates/statefulset.yaml
+++ b/kubernetes/sdnc/templates/statefulset.yaml
@@ -145,6 +145,15 @@ spec:
             name: logs
           - mountPath: {{ .Values.certpersistence.certPath }}
             name: {{ include "common.fullname" . }}-certs
+          - mountPath: {{ .Values.config.odl.salConfigDir }}/{{ .Values.config.odl.salConfigVersion}}/sal-clustering-config-{{ .Values.config.odl.salConfigVersion}}-akkaconf.xml
+            name: properties
+            subPath: akka.conf
+          - mountPath: {{ .Values.config.odl.etcDir }}/org.opendaylight.controller.cluster.datastore.cfg
+            name: properties
+            subPath: org.opendaylight.controller.cluster.datastore.cfg
+          - mountPath: {{ .Values.config.odl.binDir }}/setenv
+            name: properties
+            subPath: setenv
           resources:
 {{ include "common.resources" . | indent 12 }}
         {{- if .Values.nodeSelector }}
diff --git a/kubernetes/sdnc/values.yaml b/kubernetes/sdnc/values.yaml
index 76608d73de..dfad46ac79 100644
--- a/kubernetes/sdnc/values.yaml
+++ b/kubernetes/sdnc/values.yaml
@@ -66,6 +66,32 @@ config:
   ansiblePort: 8000
   javaHome: /usr/lib/jvm/java-1.8-openjdk
 
+  odl:
+    etcDir: /opt/opendaylight/etc
+    binDir: /opt/opendaylight/bin
+    salConfigDir: /opt/opendaylight/system/org/opendaylight/controller/sal-clustering-config
+    salConfigVersion: 1.8.2
+    akka:
+      seedNodeTimeout: 15s
+      circuitBreaker:
+        maxFailures: 10
+        callTimeout: 90s
+        resetTimeout: 30s
+      recoveryEventTimeout: 90s
+    datastore:
+      persistentActorRestartMinBackoffInSeconds: 10
+      persistentActorRestartMaxBackoffInSeconds: 40
+      persistentActorRestartResetBackoffInSeconds: 20
+      shardTransactionCommitTimeoutInSeconds: 120
+      shardIsolatedLeaderCheckIntervalInMillis: 30000
+      operationTimeoutInSeconds: 120
+    javaOptions:
+      maxGCPauseMillis: 100
+      parallelGCThreads : 3
+      numberGGLogFiles: 10
+
+
+
   #local Mariadb-galera cluster
   localDBCluster: false