Merge pull request #3927 from gchq/3909-lambda-timeout-seconds

Issue 3909 - Configure GC lambda timeout in seconds

Author: patchwork01

example/full/instance.properties

@@ -870,8 +870,8 @@ sleeper.default.partition.splitting.threshold=1000000000
 # The frequency in minutes with which the garbage collector lambda is run.
 sleeper.gc.period.minutes=15
 
-# The configurable timeout wait in minutes for the garbage collector lambda.
-sleeper.gc.lambda.timeout.minutes=14
+# The timeout in seconds for the garbage collector lambda.
+sleeper.gc.lambda.timeout.seconds=840
 
 # The amount of memory in MB for the lambda function used to perform garbage collection.
 sleeper.gc.memory.mb=4096

java/cdk/src/main/java/sleeper/cdk/stack/GarbageCollectorStack.java

@@ -53,7 +53,7 @@
 import static sleeper.core.properties.instance.GarbageCollectionProperty.GARBAGE_COLLECTOR_LAMBDA_CONCURRENCY_MAXIMUM;
 import static sleeper.core.properties.instance.GarbageCollectionProperty.GARBAGE_COLLECTOR_LAMBDA_CONCURRENCY_RESERVED;
 import static sleeper.core.properties.instance.GarbageCollectionProperty.GARBAGE_COLLECTOR_LAMBDA_MEMORY_IN_MB;
-import static sleeper.core.properties.instance.GarbageCollectionProperty.GARBAGE_COLLECTOR_LAMBDA_TIMEOUT_IN_MINUTES;
+import static sleeper.core.properties.instance.GarbageCollectionProperty.GARBAGE_COLLECTOR_LAMBDA_TIMEOUT_IN_SECONDS;
 import static sleeper.core.properties.instance.GarbageCollectionProperty.GARBAGE_COLLECTOR_PERIOD_IN_MINUTES;
 import static sleeper.core.properties.instance.GarbageCollectionProperty.GARBAGE_COLLECTOR_TABLE_BATCH_SIZE;
 import static sleeper.core.properties.instance.TableStateProperty.TABLE_BATCHING_LAMBDAS_MEMORY_IN_MB;
@@ -80,7 +80,7 @@ public GarbageCollectorStack(
         String triggerFunctionName = String.join("-", "sleeper", instanceId, "garbage-collector-trigger");
         String functionName = String.join("-", "sleeper", instanceId, "garbage-collector");
 
-        Duration handlerTimeout = Duration.seconds((60 * instanceProperties.getInt(GARBAGE_COLLECTOR_LAMBDA_TIMEOUT_IN_MINUTES)));
+        Duration handlerTimeout = Duration.seconds(instanceProperties.getInt(GARBAGE_COLLECTOR_LAMBDA_TIMEOUT_IN_SECONDS));
 
         // Garbage collector function
         IFunction triggerFunction = lambdaCode.buildFunction(this, LambdaHandler.GARBAGE_COLLECTOR_TRIGGER, "GarbageCollectorTrigger", builder -> builder

java/core/src/main/java/sleeper/core/properties/instance/BatcherProperty.java

@@ -33,6 +33,7 @@ public interface BatcherProperty {
     UserDefinedInstanceProperty INGEST_BATCHER_SUBMITTER_TIMEOUT_IN_SECONDS = Index.propertyBuilder("sleeper.ingest.batcher.submitter.timeout.seconds")
             .description("The timeout in seconds for the lambda that receives submitted requests to ingest files.")
             .defaultValue("20")
+            .validationPredicate(SleeperPropertyValueUtils::isValidLambdaTimeout)
             .propertyGroup(InstancePropertyGroup.INGEST)
             .runCdkDeployWhenChanged(true).build();
     UserDefinedInstanceProperty INGEST_BATCHER_JOB_CREATION_MEMORY_IN_MB = Index.propertyBuilder("sleeper.ingest.batcher.job.creation.memory.mb")
@@ -43,6 +44,7 @@ public interface BatcherProperty {
     UserDefinedInstanceProperty INGEST_BATCHER_JOB_CREATION_TIMEOUT_IN_SECONDS = Index.propertyBuilder("sleeper.ingest.batcher.job.creation.timeout.seconds")
             .description("The timeout in seconds for the lambda that creates ingest jobs from submitted file ingest requests.")
             .defaultValue("900")
+            .validationPredicate(SleeperPropertyValueUtils::isValidLambdaTimeout)
             .propertyGroup(InstancePropertyGroup.INGEST)
             .runCdkDeployWhenChanged(true).build();
     UserDefinedInstanceProperty INGEST_BATCHER_JOB_CREATION_LAMBDA_PERIOD_IN_MINUTES = Index.propertyBuilder("sleeper.ingest.batcher.job.creation.period.minutes")

java/core/src/main/java/sleeper/core/properties/instance/GarbageCollectionProperty.java

@@ -34,11 +34,11 @@ public interface GarbageCollectionProperty {
             .defaultValue("15")
             .propertyGroup(InstancePropertyGroup.GARBAGE_COLLECTOR)
             .runCdkDeployWhenChanged(true).build();
-    UserDefinedInstanceProperty GARBAGE_COLLECTOR_LAMBDA_TIMEOUT_IN_MINUTES = Index.propertyBuilder("sleeper.gc.lambda.timeout.minutes")
-            .description("The configurable timeout wait in minutes for the garbage collector lambda.")
-            .defaultValue("14")
+    UserDefinedInstanceProperty GARBAGE_COLLECTOR_LAMBDA_TIMEOUT_IN_SECONDS = Index.propertyBuilder("sleeper.gc.lambda.timeout.seconds")
+            .description("The timeout in seconds for the garbage collector lambda.")
+            .defaultValue("840")
+            .validationPredicate(SleeperPropertyValueUtils::isValidLambdaTimeout)
             .propertyGroup(InstancePropertyGroup.GARBAGE_COLLECTOR)
-            .validationPredicate(SleeperPropertyValueUtils::isPositiveIntegerLtEq15)
             .runCdkDeployWhenChanged(true).build();
     UserDefinedInstanceProperty GARBAGE_COLLECTOR_LAMBDA_MEMORY_IN_MB = Index.propertyBuilder("sleeper.gc.memory.mb")
             .description("The amount of memory in MB for the lambda function used to perform garbage collection.")

java/core/src/main/java/sleeper/core/properties/instance/PartitionSplittingProperty.java

@@ -56,6 +56,7 @@ public interface PartitionSplittingProperty {
     UserDefinedInstanceProperty FIND_PARTITIONS_TO_SPLIT_TIMEOUT_IN_SECONDS = Index.propertyBuilder("sleeper.partition.splitting.finder.timeout.seconds")
             .description("The timeout in seconds for the lambda function used to identify partitions that need to be split.")
             .defaultValue("900")
+            .validationPredicate(SleeperPropertyValueUtils::isValidLambdaTimeout)
             .propertyGroup(InstancePropertyGroup.PARTITION_SPLITTING)
             .runCdkDeployWhenChanged(true).build();
     UserDefinedInstanceProperty FIND_PARTITIONS_TO_SPLIT_LAMBDA_CONCURRENCY_RESERVED = Index.propertyBuilder("sleeper.partition.splitting.finder.concurrency.reserved")
@@ -76,6 +77,7 @@ public interface PartitionSplittingProperty {
     UserDefinedInstanceProperty SPLIT_PARTITIONS_TIMEOUT_IN_SECONDS = Index.propertyBuilder("sleeper.partition.splitting.timeout.seconds")
             .description("The timeout in seconds for the lambda function used to split partitions.")
             .defaultValue("900")
+            .validationPredicate(SleeperPropertyValueUtils::isValidLambdaTimeout)
             .propertyGroup(InstancePropertyGroup.PARTITION_SPLITTING)
             .runCdkDeployWhenChanged(true).build();
     UserDefinedInstanceProperty SPLIT_PARTITIONS_RESERVED_CONCURRENCY = Index.propertyBuilder("sleeper.partition.splitting.reserved.concurrency")

java/core/src/main/java/sleeper/core/properties/instance/QueryProperty.java

@@ -41,6 +41,7 @@ public interface QueryProperty {
     UserDefinedInstanceProperty QUERY_PROCESSOR_LAMBDA_TIMEOUT_IN_SECONDS = Index.propertyBuilder("sleeper.query.processor.timeout.seconds")
             .description("The timeout for the lambda that executes queries in seconds.")
             .defaultValue("900")
+            .validationPredicate(SleeperPropertyValueUtils::isValidLambdaTimeout)
             .propertyGroup(InstancePropertyGroup.QUERY)
             .runCdkDeployWhenChanged(true).build();
     UserDefinedInstanceProperty QUERY_PROCESSING_LAMBDA_STATE_REFRESHING_PERIOD_IN_SECONDS = Index.propertyBuilder("sleeper.query.processor.state.refresh.period.seconds")

java/core/src/main/java/sleeper/core/properties/instance/TableStateProperty.java

@@ -66,7 +66,7 @@ private static UserDefinedInstancePropertyImpl.Builder propertyBuilder(String pr
             .description("The timeout in seconds for lambdas that create batches of tables to run some operation against, " +
                     "eg. create compaction jobs, run garbage collection, perform partition splitting.")
             .defaultValue("60")
-            .validationPredicate(SleeperPropertyValueUtils::isPositiveInteger)
+            .validationPredicate(SleeperPropertyValueUtils::isValidLambdaTimeout)
             .propertyGroup(InstancePropertyGroup.TABLE_STATE)
             .build();
     UserDefinedInstanceProperty TABLE_PROPERTIES_PROVIDER_TIMEOUT_IN_MINS = Index.propertyBuilder("sleeper.cache.table.properties.provider.timeout.minutes")
@@ -112,7 +112,7 @@ private static UserDefinedInstancePropertyImpl.Builder propertyBuilder(String pr
     UserDefinedInstanceProperty SNAPSHOT_CREATION_LAMBDA_TIMEOUT_IN_SECONDS = Index.propertyBuilder("sleeper.statestore.snapshot.creation.lambda.timeout.seconds")
             .description("The timeout in seconds after which to terminate the transaction log snapshot creation lambda.")
             .defaultValue("900")
-            .validationPredicate(SleeperPropertyValueUtils::isPositiveInteger)
+            .validationPredicate(SleeperPropertyValueUtils::isValidLambdaTimeout)
             .propertyGroup(InstancePropertyGroup.TABLE_STATE)
             .build();
     UserDefinedInstanceProperty SNAPSHOT_CREATION_LAMBDA_MEMORY = Index.propertyBuilder("sleeper.statestore.snapshot.creation.memory.mb")
@@ -178,7 +178,7 @@ private static UserDefinedInstancePropertyImpl.Builder propertyBuilder(String pr
             .propertyGroup(InstancePropertyGroup.TABLE_STATE).build();
     UserDefinedInstanceProperty TRANSACTION_DELETION_LAMBDA_TIMEOUT_SECS = Index.propertyBuilder("sleeper.statestore.transaction.deletion.lambda.timeout.seconds")
             .description("The maximum timeout for the transaction deletion lambda in seconds.")
-            .validationPredicate(SleeperPropertyValueUtils::isPositiveInteger)
+            .validationPredicate(SleeperPropertyValueUtils::isValidLambdaTimeout)
             .defaultValue("900")
             .propertyGroup(InstancePropertyGroup.TABLE_STATE).build();
     UserDefinedInstanceProperty TABLE_INDEX_DYNAMO_POINT_IN_TIME_RECOVERY = Index.propertyBuilder("sleeper.tables.index.dynamo.pointintimerecovery")

java/core/src/main/java/sleeper/core/properties/validation/SleeperPropertyValueUtils.java

@@ -77,17 +77,6 @@ public static boolean isPositiveIntegerLtEq10(String integer) {
         return parseAndCheckInteger(integer, num -> num > 0 && num <= 10);
     }
 
-    /**
-     * Checks if a property value is a whole positive number of minutes within the maximum timeout for an invocation of
-     * AWS Lambda.
-     *
-     * @param  integer the value
-     * @return         true if the value meets the requirement
-     */
-    public static boolean isPositiveIntegerLtEq15(String integer) {
-        return parseAndCheckInteger(integer, num -> num > 0 && num <= 15);
-    }
-
     /**
      * Checks if a property value is an integer greater than or equal to 0.
      *

java/core/src/test/java/sleeper/core/properties/instance/InstancePropertiesTest.java

@@ -80,7 +80,7 @@
 import static sleeper.core.properties.instance.CompactionProperty.MAXIMUM_CONCURRENT_COMPACTION_TASKS;
 import static sleeper.core.properties.instance.GarbageCollectionProperty.GARBAGE_COLLECTOR_LAMBDA_CONCURRENCY_RESERVED;
 import static sleeper.core.properties.instance.GarbageCollectionProperty.GARBAGE_COLLECTOR_LAMBDA_MEMORY_IN_MB;
-import static sleeper.core.properties.instance.GarbageCollectionProperty.GARBAGE_COLLECTOR_LAMBDA_TIMEOUT_IN_MINUTES;
+import static sleeper.core.properties.instance.GarbageCollectionProperty.GARBAGE_COLLECTOR_LAMBDA_TIMEOUT_IN_SECONDS;
 import static sleeper.core.properties.instance.GarbageCollectionProperty.GARBAGE_COLLECTOR_PERIOD_IN_MINUTES;
 import static sleeper.core.properties.instance.IngestProperty.ECR_INGEST_REPO;
 import static sleeper.core.properties.instance.IngestProperty.INGEST_KEEP_ALIVE_PERIOD_IN_SECONDS;
@@ -268,7 +268,7 @@ private static InstanceProperties getSleeperProperties() {
         instanceProperties.set(VPC_ID, "aVPC");
         instanceProperties.set(SUBNETS, "subnet1");
         instanceProperties.setNumber(GARBAGE_COLLECTOR_PERIOD_IN_MINUTES, 20);
-        instanceProperties.setNumber(GARBAGE_COLLECTOR_LAMBDA_TIMEOUT_IN_MINUTES, 15);
+        instanceProperties.setNumber(GARBAGE_COLLECTOR_LAMBDA_TIMEOUT_IN_SECONDS, 840);
         instanceProperties.setNumber(QUEUE_VISIBILITY_TIMEOUT_IN_SECONDS, 600);
         instanceProperties.setNumber(COMPACTION_KEEP_ALIVE_PERIOD_IN_SECONDS, 700);
         instanceProperties.setNumber(INGEST_KEEP_ALIVE_PERIOD_IN_SECONDS, 800);

scripts/templates/instanceproperties.template

@@ -899,8 +899,8 @@ sleeper.default.partition.splitting.threshold=1000000000
 # The frequency in minutes with which the garbage collector lambda is run.
 sleeper.gc.period.minutes=15
 
-# The configurable timeout wait in minutes for the garbage collector lambda.
-sleeper.gc.lambda.timeout.minutes=14
+# The timeout in seconds for the garbage collector lambda.
+sleeper.gc.lambda.timeout.seconds=840
 
 # The amount of memory in MB for the lambda function used to perform garbage collection.
 sleeper.gc.memory.mb=4096