RESTful Quarkus serverless cold start times are very slow on AWS

[zkhan96]

Describe the bug I seem to have cold start times of 11s when doing any rest call to my quarkus app in AWS lambda deployed by sam deploy

The app is using API gateway as a proxy to talk to an endpoint on the app.

The app connects to RDS PostgreSQL using a custom credentials provider, does a GET on a very small payload, and returns the JSON to the client.

Expected behaviour Cold start times of 1s.

To Reproduce I am running the quarkus app in JVM mode and building the uber jar by doing:

• ./gradlew quarkusBuild --uber-jar
• sam deploy

Configuration

  quarkus:
    datasource:
      credentials-provider: custom
      credentials-provider-name: pg-credentials-provider
      jdbc:
        url: jdbc:postgresql://some-cluster-name
  custom:
    host: some-rds-host-for-custom-credentials-provider

Screenshots

A GET on the /health endpoint using the quarkus-smallrye-health extension

Environment (please complete the following information):

• Output of java -version:

  openjdk version "11.0.4" 2019-07-16
  OpenJDK Runtime Environment AdoptOpenJDK (build 11.0.4+11)
  OpenJDK 64-Bit Server VM AdoptOpenJDK (build 11.0.4+11, mixed mode)

• GraalVM version (if different from Java): Not used (yet)
• Quarkus version or git rev: 1.5.2.Final
• Build tool (ie. output of mvnw --version or gradlew --version):

  
  ------------------------------------------------------------
  Gradle 6.5
  ------------------------------------------------------------

Build time: 2020-06-02 20:46:21 UTC Revision: a27f41e4ae5e8a41ab9b19f8dd6d86d7b384dad4

Kotlin: 1.3.72 Groovy: 2.5.11 Ant: Apache Ant(TM) version 1.10.7 compiled on September 1 2019 JVM: 11.0.4 (AdoptOpenJDK 11.0.4+11) OS: Mac OS X 10.14.6 x86_64

**Additional context**
Here are all my dependencies from my build.gradle:

implementation enforcedPlatform("${quarkusPlatformGroupId}:${quarkusPlatformArtifactId}:${quarkusPlatformVersion}")
compileOnly 'org.projectlombok:lombok:1.18.12'
annotationProcessor 'org.projectlombok:lombok:1.18.12'
implementation 'org.lorislab.quarkus:quarkus-hibernate-types:0.1.2'
implementation 'io.quarkus:quarkus-config-yaml'
implementation 'io.quarkus:quarkus-amazon-lambda-http'
implementation 'io.quarkus:quarkus-resteasy'
implementation 'io.quarkus:quarkus-resteasy-jackson'
implementation 'io.quarkus:quarkus-jdbc-postgresql'
implementation 'io.quarkus:quarkus-hibernate-orm'
implementation 'io.quarkus:quarkus-logging-json'
implementation 'io.quarkus:quarkus-spring-data-jpa'
implementation 'io.quarkus:quarkus-smallrye-health'
implementation group: 'com.amazonaws', name: 'aws-java-sdk-rds', version: '1.11.804'

testCompile 'org.testcontainers:junit-jupiter:1.12.5'
testCompile 'org.testcontainers:postgresql:1.12.5'
testImplementation 'io.quarkus:quarkus-junit5'
testImplementation 'io.rest-assured:rest-assured'
testCompile 'io.quarkus:quarkus-junit5-mockito'
testCompileOnly 'org.projectlombok:lombok:1.18.12'
testAnnotationProcessor 'org.projectlombok:lombok:1.18.12'
testCompile 'io.quarkus:quarkus-flyway'

Here is the cloudwatch log:

Duration: 5461.09 ms Billed Duration: 5500 ms Memory Size: 512 MB Max Memory Used: 220 MB Init Duration: 5009.05 ms

[martinoneutrino]

What is the performance with the lambda given a 2048MB memory allocation? Lambda CPU is proporational to memory assigned, and with millisecond billing now, it makes sense to increase the memory to get a nearly proportionally faster response time.

[geoand]

Is this still an issue with the latest Quarkus versions?

[geoand]

Closing for lack of feedback.