Degressly

---

Degressly is a portmanteau of "Deterministic Regression". Inspired by opendiffy/diffy, degressly aims to create a scalable framework for performing regression testing.

Degressly works by running three parallel instances of your code side by side and multicasting all inbound requests to all three instances. The primary and secondary instances run your last known good code, while the candidate instances runs code that is to be tested. Differences between primary and candidate instances involve noise from non-deterministic sources like random number generation and timestamps, these differences are ignored based on the differences obtained from responses of primary and secondary instances.

If you want a quick and dirty way to try out degressly with a sample application, check out degressly/degressly-demo.

The degressly ecosystem depends on the following repositories:

Repository	Description
degressly-core	Core frontend HTTP Proxy. Logs responses from each downstream separately, can be configured to push observations for analysis by degressly-comparator.
degressly-comparator	Analyzes observations received from Primary, Secondary and Candidate deployments
degressly-downstream	Downstream proxy for services that make S2S calls to downstream services, with the ability to accomodate non-idempotent downstreams. Logs requests from each downstream separately, can be configured to push observations for analysis by degressly-comparator.

Getting Started

Clone repository:

git clone https://github.com/degressly/degressly-core.git
cd degressly-core

Running Docker singleton

docker build -f Dockerfile -t degressly-core:latest
docker run -p8000:8000 degressly-core:latest

Running with docker-compose

Create .env file:

touch .env
nano .env

Sample .env file:

primary_host=http://host.docker.internal:9000
secondary_host=http://host.docker.internal:9001
candidate_host=http://host.docker.internal:9002

Note on host names: https://docs.docker.com/desktop/networking/#i-want-to-connect-from-a-container-to-a-service-on-the-host

docker build -f Dockerfile -t degressly-core:latest
docker compose up

Running full degressly ecosystem with docker-compose:

Clone all repos and build docker images:

git clone https://github.com/degressly/degressly-core.git
git clone https://github.com/degressly/degressly-comparator.git
git clone https://github.com/degressly/degressly-downstream.git
docker build degressly-core/ -t degressly-core:latest 
docker build degressly-comparator/ -t degressly-comparator:latest 
docker build degressly-downstream/ -t degressly-downstream:latest 
cd degressly-core

Create .env file:

touch .env

spring_profiles_active=mongo
diff_publisher_bootstrap_servers=kafka:9092
diff_publisher_topic_name=diff_stream
primary_host=http://host.docker.internal:9000
secondary_host=http://host.docker.internal:9001
candidate_host=http://host.docker.internal:9002
MONGO_URL=<mongo_cluster_url>/<mongo_db_name>
MONGO_USERNAME=<mongo_username>
MONGO_PASSWORD=<mongo_password>
MONGO_DBNAME=<mongo_db_name>

Note on host names: https://docs.docker.com/desktop/networking/#i-want-to-connect-from-a-container-to-a-service-on-the-host

Run containers

docker compose --profile full up

Config flags

VM Options (When running jar)	Environment Variables (When using Docker)	Example	Description
diff.publisher.bootstrap-servers	diff_publisher_bootstrap_servers	false(default) | kafka:9092	Address of kafka bootstrap servers for integration with degressly-comparator
diff.publisher.topic-name	diff_publisher_topic_name	diff_stream	Kafka topic name for integration with degressly-comparator
primary.host	primary_host	http://localhost:9000 / http://host.docker.internal:9000	Forwarding address of primary instance
secondary.host	secondary_host	http://localhost:9001 / http://host.docker.internal:9001	Forwarding address of secondary instance
candidate.host	candidate_host	http://localhost:9002 / http://host.docker.internal:9002	Forwarding address of candidate instance
return.response.from	return_response_from	PRIMARY(default) | SECONDARY | CANDIDATE	Which instance's response is to be returned to the user.
wait.after.forwarding.to.primary	wait_after_forwarding_to_primary	100	Time to wait(in ms) after sending the request to primary and before sending it to secondary and candidate replicas. May be negative in which case request will be sent to primary and candidate instances before primary instance.

Limitations / TODO

In no particular order:

• DB layer observation recon (comparator can consume from debezium or similar CDC pipeline).
• DB Proxy...?
• Performance regression tracking.
• Dockerization.
• Documentation for Replay support.

Support

If you would like to reach out for support or feature requests, feel free to drop an email at [email protected]