conductor-community icon indicating copy to clipboard operation
conductor-community copied to clipboard

new version can't be start with postgres

Open wyaopeng opened this issue 2 years ago • 3 comments

Describe the bug new version can't be start with postgres,

image

can you tell me ,how to fix it ?

wyaopeng avatar Aug 26 '22 10:08 wyaopeng

i solved a similar issue adding a runtime dependency (version may vary) into build.gradle :

runtimeOnly "com.netflix.conductor:conductor-postgres-persistence:3.11.3"

still i have a problem, a fresh install with the dependency works just one time. After a restart i have problems:

org.springframework.beans.factory.UnsatisfiedDependencyException: Error creating bean with name 'taskResource' defined in URL [jar:file:/app/libs/conductor-server-3.11.3-boot.jar ... Caused by: org.springframework.beans.factory.UnsatisfiedDependencyException: Error creating bean with name 'taskServiceImpl' ... Caused by: org.springframework.beans.factory.UnsatisfiedDependencyException: Error creating bean with name 'executionService' ... Caused by: org.springframework.beans.factory.UnsatisfiedDependencyException: Error creating bean with name 'executionDAOFacade' ... Error creating bean with name 'es6IndexRestDAO' ... Caused by: java.net.SocketTimeoutException: 30,000 milliseconds timeout on connection http-outgoing-0

ps. i start conductor with this command: docker compose -f docker-compose.yaml -f docker-compose-postgres.yaml up

any suggestions?

@wyaopeng @jxu-nflx

thx

e.

edoang avatar Sep 27 '22 19:09 edoang

Yea, it seems the server can't reach the ES6 service/container. The config-postgres.properties config sets the ES6 endpoint to conductor.elasticsearch.url=http://es:9200. The seems to be in line with the Compose setup.

yuri1969 avatar Dec 28 '22 14:12 yuri1969

@edoang Enabling the DEBUG logs it can be seen that the ES6 timeout gets hit because the Conductor server is waiting for the ES6 cluster to get ready (be in the green state). The cluster can't get fully ready since the Docker Compose configures a single node cluster with the default count of replicas.

However, you can apply a workaround by adding conductor.elasticsearch.clusterHealthColor=yellow to the config-postgres.properties, rebuild, and 🎉

Actually, this workaround can be found in the official doc under the Elasticsearch timeout FAQ section.

yuri1969 avatar Dec 28 '22 22:12 yuri1969