Skip to main content

Celery Connectors

Project description

Celery Connectors

Celery is a great framework for processing messages from a message queue broker like Redis or RabbitMQ. If you have a queue with json or pickled messages that you need to consume and process, then hopefully this repository will help you out.

It has multiple examples on setting up working publisher-subscriber messaging workflows using Celery, Celery Bootsteps, Kombu, and Kombu mixins. These examples are focused on finding a starting ground to tune for high availability + performance + reduce the risk of message loss (the dockerized celery bootstep rabbitmq subscriber can process around 100,000 messages in 90 seconds with 3 workers). By using the included docker containers combined with the included load tests, you can start to vet your solution won’t wake you up in the middle of the night during an outage.

Each example below can run as a docker container with the included docker-compose files in the compose directory. Please note these docker-compose steps are optional and the consumer counts in the documentation below will only refer to the non-dockerized, repository versions.

Here’s the JSON-to-Celery ecomm relay example in action. By using docker-compose you can use container monitoring tools to benchmark resources and throughput to figure out your deployment footprint and address bottlenecks.

https://raw.githubusercontent.com/jay-johnson/celery-connectors/master/_images/celery-connectors-json-to-celery-relay-with-existing-ecomm-celery-app.gif

Why do I care?

  • Do you want to read json or pickled messages out of a queue and have a framework handle the scaling and deployment aspects all out of the box?
  • Do you want a simple way to read out of queues without setting up a task result backend (mongo)?
  • Do you want to connect a windows python client to a backend linux system or cluster?
  • Do you want to communicate with all your AWS VPC backends over SQS?
  • Do you want to glue python and non-python technologies together through a message queue backend?
  • Do you want something that works with python 2 and 3?

How do I get started?

  1. Setup the virtualenv

    If you want to use python 2:

    virtualenv venv && source venv/bin/activate && pip install celery-connectors
    

    If you want to use python 3:

    virtualenv -p python3 venv && source venv/bin/activate && pip install celery-connectors
    
  2. Confirm the pip is installed

    pip list | grep celery-connectors
    
  3. Start the containers

    # if you do not have docker compose installed, you can try installing it with:
    # pip install docker-compose
    start-redis-and-rabbitmq.sh
    

    Or if your docker version and OS support container volume-mounting, then you can persist Redis and RabbitMQ messages and data to disk with:

    ./start-persistence-containers.sh
    
  4. Check the Redis and RabbitMQ containers are running

    docker ps
    CONTAINER ID        IMAGE                       COMMAND                  CREATED             STATUS              PORTS                                                                                                       NAMES
    913e8092dbde        mher/flower:latest          "/usr/local/bin/py..."   35 seconds ago      Up 35 seconds                                                                                                                   celflowerredis
    b6983a1316ba        rabbitmq:3.6.6-management   "docker-entrypoint..."   35 seconds ago      Up 34 seconds       4369/tcp, 5671/tcp, 0.0.0.0:5672->5672/tcp, 0.0.0.0:15672->15672/tcp, 15671/tcp, 0.0.0.0:25672->25672/tcp   celrabbit1
    52cb4c511d61        redis:4.0.5-alpine          "docker-entrypoint..."   35 seconds ago      Up 34 seconds       0.0.0.0:6379->6379/tcp, 0.0.0.0:16379->16379/tcp                                                            celredis1
    202bdaf70784        mher/flower:latest          "/usr/local/bin/py..."   35 seconds ago      Up 35 seconds                                                                                                                   celflowerrabbit
    

Running a Payments JSON-to-JSON Relay Service

This will simulate a json->json relay using kombu mixins:

http://docs.celeryproject.org/projects/kombu/en/latest/reference/kombu.mixins.html

Kombu mixins are a great way to process messages without Celery, and they are resilient to multiple HA scenarios including a complete broker failures. While building this I would load up messages to process, simulate lag before an ack and then start/stop the RabbitMQ docker container to see how things reacted. As long as the subscribers can declare their consuming queues on a fresh broker start-up case, these mixins seem capable of surviving these types of DR events. By default these builds are going to only read one message out of the queue at a time.

Start JSON Relay

This process will consume JSON dictionary messages on the ecomm.api.west RabbitMQ queue and pass the message to the reporting.payments queue.

Please start this in a new terminal that has sourced the virtual env: source venv/bin/activate

start-mixin-json-relay.py
INFO:mixin_relay:Consuming queues=1
INFO:relay:consuming queues=[<unbound Queue ecomm.api.west -> <unbound Exchange ecomm.api(topic)> -> ecomm.api.west>]
INFO:kombu.mixins:Connected to amqp://rabbitmq:**@127.0.0.1:5672//
INFO:relay-wrk:creating consumer for queues=1 callback=handle_message relay_ex=Exchange ''(direct) relay_rk=reporting.payments prefetch=1

Or with docker compose

docker-compose -f compose-start-mixin-json-relay.yml up
Starting jtojrelay ...
Starting jtojrelay ... done
Attaching to jtojrelay
jtojrelay    | 2017-12-15 06:37:39,458 - jtoj_relay - INFO - Consuming queues=1
jtojrelay    | 2017-12-15 06:37:39,462 - jtoj_relay - INFO - consuming queues=[<unbound Queue ecomm.api.west -> <unbound Exchange ""(topic)> -> ecomm.api.west>]
jtojrelay    | 2017-12-15 06:37:39,478 - kombu.mixins - INFO - Connected to amqp://rabbitmq:**@127.0.0.1:5672//

List the Queues

In a new terminal that has the virtual env loaded, checkout the RabbitMQ queues:

list-queues.sh

Listing Queues broker=localhost:15672
name consumers messages messages_ready messages_unacknowledged
celeryev.ea44162e-7224-4167-be30-4be614c33fc9 1 0 0 0
ecomm.api.west 1 0 0 0

Start the Kombu Mixin Subscriber

In a new terminal that has the virtual env loaded, start the subscriber for relayed messags in the reporting.payments queue:

kombu_mixin_subscriber.py
INFO:kombu-mixin-subscriber:Start - kombu-mixin-subscriber
INFO:kombu-subscriber:setup routing
INFO:kombu-subscriber:kombu-mixin-subscriber - kombu.subscriber queues=reporting.payments consuming with callback=handle_message

Or with docker compose:

docker-compose -f compose-kombu-mixin-subscriber.yml  up
WARNING: Found orphan containers (jtojrelay) for this project. If you removed or renamed this service in your compose file, you can run this command with the --remove-orphans flag to clean it up.
Creating kombumixinsubrmq ... done
Attaching to kombumixinsubrmq
kombumixinsubrmq    | 2017-12-15 06:41:15,135 - kombu-mixin-subscriber - INFO - Start - kombu-mixin-subscriber
kombumixinsubrmq    | 2017-12-15 06:41:15,135 - kombu-subscriber - INFO - setup routing

List the Bindings

With the relay and the subscrbier online the bindings should show two separate queues for these two processes.

list-bindings.sh

Listing Bindings broker=localhost:15672
source destination routing_key
  celeryev.ea44162e-7224-4167-be30-4be614c33fc9 celeryev.ea44162e-7224-4167-be30-4be614c33fc9
  ecomm.api.west ecomm.api.west
  reporting.payments reporting.payments
celeryev celeryev.ea44162e-7224-4167-be30-4be614c33fc9 #
ecomm.api ecomm.api.west ecomm.api.west
reporting.payments reporting.payments reporting.payments

Publish Ecomm messages to the Relay

In a new terminal that has the virtual env loaded, start the mixin publisher that will send JSON messages to the ecomm.api.west queue:

start-mixin-publisher.py
INFO:robopub:Generating messages=10
INFO:robopub:Publishing messages=10
INFO:pub_send:pub_send publish - ex=Exchange ecomm.api(topic) rk=ecomm.api.west sz=json
INFO:pub_send:pub_send publish - ex=Exchange ecomm.api(topic) rk=ecomm.api.west sz=json
INFO:pub_send:pub_send publish - ex=Exchange ecomm.api(topic) rk=ecomm.api.west sz=json
INFO:pub_send:pub_send publish - ex=Exchange ecomm.api(topic) rk=ecomm.api.west sz=json
INFO:pub_send:pub_send publish - ex=Exchange ecomm.api(topic) rk=ecomm.api.west sz=json
INFO:pub_send:pub_send publish - ex=Exchange ecomm.api(topic) rk=ecomm.api.west sz=json
INFO:pub_send:pub_send publish - ex=Exchange ecomm.api(topic) rk=ecomm.api.west sz=json
INFO:pub_send:pub_send publish - ex=Exchange ecomm.api(topic) rk=ecomm.api.west sz=json
INFO:pub_send:pub_send publish - ex=Exchange ecomm.api(topic) rk=ecomm.api.west sz=json
INFO:pub_send:pub_send publish - ex=Exchange ecomm.api(topic) rk=ecomm.api.west sz=json
INFO:robopub:Done Publishing

Or with docker compose:

docker-compose -f compose-start-mixin-publisher.yml up
WARNING: Found orphan containers (kombumixinsubrmq, jtojrelay) for this project. If you removed or renamed this service in your compose file, you can run this command with the --remove-orphans flag to clean it up.
Starting mixinpubrmq ... done
Attaching to mixinpubrmq

Verify the Relay Handled the Messages

Verify the terminal logs in the relay look similar to:

INFO:relay-wrk:default handle_message - acking - msg={'data': {'simulated_lag': 1.0}, 'msg_id': '35e8546f-f757-4764-9a25-12b867f61957_1', 'created': '2017-12-13T01:30:35.401399'}
INFO:relay-wrk:send start - relay_ex=Exchange ''(direct) relay_rk=reporting.payments id=95c93115-2041-424b-b37e-0e8dff1b6336_1
INFO:pub_send:pub_send publish - ex=Exchange ''(direct) rk=reporting.payments sz=json
INFO:relay-wrk:send done - id=95c93115-2041-424b-b37e-0e8dff1b6336_1
INFO:relay-wrk:default handle_message - acking - msg={'data': {'simulated_lag': 1.0}, 'msg_id': '989641cc-cd2b-4041-81aa-bdd27393646a_1', 'created': '2017-12-13T01:30:35.401529'}
INFO:relay-wrk:send start - relay_ex=Exchange ''(direct) relay_rk=reporting.payments id=7d8b473a-1f7e-4d04-8e8a-234536b0a8fb_1
INFO:pub_send:pub_send publish - ex=Exchange ''(direct) rk=reporting.payments sz=json
INFO:relay-wrk:send done - id=7d8b473a-1f7e-4d04-8e8a-234536b0a8fb_1
INFO:relay-wrk:default handle_message - acking - msg={'data': {'simulated_lag': 1.0}, 'msg_id': '68eb6ab0-2e41-4838-a088-927709c4d595_1', 'created': '2017-12-13T01:30:35.401554'}
INFO:relay-wrk:send start - relay_ex=Exchange ''(direct) relay_rk=reporting.payments id=4ca34760-db69-4c06-97c9-0355c38bd158_1
INFO:pub_send:pub_send publish - ex=Exchange ''(direct) rk=reporting.payments sz=json
INFO:relay-wrk:send done - id=4ca34760-db69-4c06-97c9-0355c38bd158_1
INFO:relay-wrk:default handle_message - acking - msg={'data': {'simulated_lag': 1.0}, 'msg_id': 'f906ab52-27f1-4ea7-bd68-2956da232258_1', 'created': '2017-12-13T01:30:35.401618'}
INFO:relay-wrk:send start - relay_ex=Exchange ''(direct) relay_rk=reporting.payments id=8a584a99-b35d-4e18-acd8-45d32871ba0a_1

Verify the Subscriber Handled the Relayed Messages

INFO:kombu-mixin-subscriber:callback received msg body={'msg_id': '95c93115-2041-424b-b37e-0e8dff1b6336_1', 'data': {'org_msg': {'msg_id': '35e8546f-f757-4764-9a25-12b867f61957_1', 'data': {'simulated_lag': 1.0}, 'created': '2017-12-13T01:30:35.401399'}, 'relay_name': 'json-to-json-relay'}, 'created': '2017-12-13T01:30:35.423314'}
INFO:kombu-subscriber:kombu-mixin-subscriber - kombu.subscriber queues=reporting.payments consuming with callback=handle_message
INFO:kombu-mixin-subscriber:callback received msg body={'msg_id': '7d8b473a-1f7e-4d04-8e8a-234536b0a8fb_1', 'data': {'org_msg': {'msg_id': '989641cc-cd2b-4041-81aa-bdd27393646a_1', 'data': {'simulated_lag': 1.0}, 'created': '2017-12-13T01:30:35.401529'}, 'relay_name': 'json-to-json-relay'}, 'created': '2017-12-13T01:30:35.445645'}
INFO:kombu-subscriber:kombu-mixin-subscriber - kombu.subscriber queues=reporting.payments consuming with callback=handle_message
INFO:kombu-mixin-subscriber:callback received msg body={'msg_id': '4ca34760-db69-4c06-97c9-0355c38bd158_1', 'data': {'org_msg': {'msg_id': '68eb6ab0-2e41-4838-a088-927709c4d595_1', 'data': {'simulated_lag': 1.0}, 'created': '2017-12-13T01:30:35.401554'}, 'relay_name': 'json-to-json-relay'}, 'created': '2017-12-13T01:30:35.453077'}
INFO:kombu-subscriber:kombu-mixin-subscriber - kombu.subscriber queues=reporting.payments consuming with callback=handle_message
INFO:kombu-mixin-subscriber:callback received msg body={'msg_id': '8a584a99-b35d-4e18-acd8-45d32871ba0a_1', 'data': {'org_msg': {'msg_id': 'f906ab52-27f1-4ea7-bd68-2956da232258_1', 'data': {'simulated_lag': 1.0}, 'created': '2017-12-13T01:30:35.401618'}, 'relay_name': 'json-to-json-relay'}, 'created': '2017-12-13T01:30:35.458601'}

Confirm the Queues are empty

list-queues.sh

Listing Queues broker=localhost:15672
name consumers messages messages_ready messages_unacknowledged
celeryev.ea44162e-7224-4167-be30-4be614c33fc9 1 0 0 0
ecomm.api.west 1 0 0 0
reporting.payments 1 0 0 0

Stop the the JSON Relay Demo

In the mixin relay and mixin subscriber terminal sessions use: ctrl + c to stop the processes.

Restart the docker containers to a good, clean state.

Stop:

stop-redis-and-rabbitmq.sh
Stopping redis and rabbitmq
Stopping celrabbit1      ... done
Stopping celredis1       ... done
Stopping celflowerredis  ... done
Stopping celflowerrabbit ... done

Start:

start-redis-and-rabbitmq.sh
Starting redis and rabbitmq
Creating celrabbit1 ... done
Creating celrabbit1 ... done
Creating celredis1 ... done
Creating celflowerredis ... done

Running an Ecommerce JSON-to-Celery Relay Service

This will simulate hooking up an existing Celery application to start processing Celery tasks from JSON messages in a RabbitMQ queue. This is useful because it allows reusing existing Celery application tasks over a JSON messaging layer for mapping payloads to specific, existing Celery tasks. With this approach you can glue python and non-python services together provided that they can publish JSON messages to Redis, RabbitMQ or AWS SQS (please refer to the fix SQS section). Each of the components below can scale horizontally for redundancy. Each one also utilizes native RabbitMQ acks (https://www.rabbitmq.com/confirms.html) to ensure messages are never deleted or lost until propagation to the next queue or component has been confirmed.

Note: Please run this demo with three separate terminal sessions and a browser to view the Celery application’s task progress in Flower.

Start Ecommerce Celery Worker

Start a Celery worker for an existing ecommerce application from a hypothetical Django or Flask server.

Note: Please run this from the base directory for the repository and source the virtual env: source venv/bin/activate

./start-ecomm-worker.sh

-------------- celery@ecommerce_subscriber v4.1.0 (latentcall)
---- **** -----
--- * ***  * -- Linux-4.7.4-200.fc24.x86_64-x86_64-with-fedora-24-Twenty_Four 2017-12-14 00:33:02
-- * - **** ---
- ** ---------- [config]
- ** ---------- .> app:         ecommerce-worker:0x7f0c23f1c550
- ** ---------- .> transport:   amqp://rabbitmq:**@localhost:5672//
- ** ---------- .> results:     redis://localhost:6379/10
- *** --- * --- .> concurrency: 3 (prefork)
-- ******* ---- .> task events: OFF (enable -E to monitor tasks in this worker)
--- ***** -----
-------------- [queues]
                .> celery           exchange=celery(direct) key=celery


[tasks]
. ecomm_app.ecommerce.tasks.handle_user_conversion_events

[2017-12-14 00:33:02,243: INFO/MainProcess] Connected to amqp://rabbitmq:**@127.0.0.1:5672//
[2017-12-14 00:33:02,260: INFO/MainProcess] mingle: searching for neighbors
[2017-12-14 00:33:03,293: INFO/MainProcess] mingle: all alone
[2017-12-14 00:33:03,337: INFO/MainProcess] celery@ecommerce_subscriber ready.
[2017-12-14 00:33:05,275: INFO/MainProcess] Events of group {task} enabled by remote.

Or with docker compose:

docker-compose -f compose-start-ecomm-worker.yml up
Recreating ecommworker ... done
Attaching to ecommworker

Notice the worker is named celery@ecommerce_subscriber this is the identifier for viewing the Celery application in Flower:

http://localhost:5555/worker/celery@ecommerce_subscriber (login: admin/admin)

Start Ecomm Relay

This process will consume JSON dictionary messages on the ecomm.api.west RabbitMQ queue and pass the message to the ecomm Celery app as a ecomm_app.ecommerce.tasks.handle_user_conversion_events Celery task.

Please start this in a new terminal that has sourced the virtual env: source venv/bin/activate

./start-mixin-celery-relay.py
2017-12-14 00:36:47,339 - jtoc_relay - INFO - Consuming queues=1
2017-12-14 00:36:47,342 - jtoc - INFO - consuming queues=[<unbound Queue ecomm.api.west -> <unbound Exchange ecomm.api(topic)> -> ecomm.api.west>]
2017-12-14 00:36:47,353 - kombu.mixins - INFO - Connected to amqp://rabbitmq:**@127.0.0.1:5672//
2017-12-14 00:36:47,355 - jtoc - INFO - creating consumer for queues=1 callback=handle_message relay_ex=Exchange ''(direct) relay_rk=reporting.payments prefetch=1

Or with docker compose:

docker-compose -f compose-start-mixin-celery-relay.yml up
Creating jtocrelay ... done
Attaching to jtocrelay
jtocrelay    | 2017-12-15 06:56:07,689 - jtoc_relay - INFO - Consuming queues=1
jtocrelay    | 2017-12-15 06:56:07,703 - jtoc_relay - INFO - consuming queues=[<unbound Queue ecomm.api.west -> <unbound Exchange ecomm.api(topic)> -> ecomm.api.west>]
jtocrelay    | 2017-12-15 06:56:07,720 - kombu.mixins - INFO - Connected to amqp://rabbitmq:**@127.0.0.1:5672//

Publish a User Conversion Event to the Ecomm Relay

This will use Kombu to publish a JSON dictionary message to the ecomm.api.west RabbitMQ queue which is monitored by the mixin JSON to Celery relay. This test tool is configured to simulate hypothetical worst-cast lag during the relay + message processing. This is a functional test to ensure everything stays connected and ready for more messages to process.

Please start this in a new terminal that has sourced the virtual env: source venv/bin/activate

./start-mixin-publisher.py
2017-12-14 00:42:16,849 - robopub - INFO - Generating messages=10
2017-12-14 00:42:16,850 - robopub - INFO - Publishing messages=10
2017-12-14 00:42:16,866 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=46cb24f0d0_1
2017-12-14 00:42:16,867 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=d2724b75fa_1
2017-12-14 00:42:16,867 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=e72e09da34_1
2017-12-14 00:42:16,869 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=f5ec3f0c9d_1
2017-12-14 00:42:16,870 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=222094db10_1
2017-12-14 00:42:16,871 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=9bed4cc0e5_1
2017-12-14 00:42:16,871 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=f66139a9cf_1
2017-12-14 00:42:16,872 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=94d3a2c7ed_1
2017-12-14 00:42:16,873 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=b517f87ff4_1
2017-12-14 00:42:16,873 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=822ef4142c_1
2017-12-14 00:42:16,874 - robopub - INFO - Done Publishing

Or with docker compose:

docker-compose -f compose-start-mixin-publisher.yml up
WARNING: Found orphan containers (jtocrelay) for this project. If you removed or renamed this service in your compose file, you can run this command with the --remove-orphans flag to clean it up.
Recreating mixinpubrmq ... done
Attaching to mixinpubrmq
mixinpubrmq    | 2017-12-15 06:56:43,517 - robopub - INFO - Generating messages=10

Verify the Ecomm Relay Processed the Conversion Message

After the simulated lag finishes, the logs for the ecomm relay should show something similar to:

2017-12-14 00:42:16,869 - jtoc - INFO - hd msg=46cb24f0d0_1 from_ex=ecomm.api from_rk=ecomm.api.west
2017-12-14 00:42:16,870 - jtoc - INFO - relay msg_id=46cb24f0d0_1 body={'msg_id': '46cb24f0d0_1', 've
2017-12-14 00:42:16,937 - jtoc - INFO - relay done with msg_id=46cb24f0d0_1
2017-12-14 00:42:16,937 - jtoc - INFO - task - ecomm_app.ecommerce.tasks.handle_user_conversion_events - simulating processing lag sleep=8.0 seconds
2017-12-14 00:42:24,947 - jtoc - INFO - hd msg=d2724b75fa_1 from_ex=ecomm.api from_rk=ecomm.api.west
2017-12-14 00:42:24,947 - jtoc - INFO - relay msg_id=d2724b75fa_1 body={'msg_id': 'd2724b75fa_1', 've
2017-12-14 00:42:24,953 - jtoc - INFO - relay done with msg_id=d2724b75fa_1
2017-12-14 00:42:24,953 - jtoc - INFO - task - ecomm_app.ecommerce.tasks.handle_user_conversion_events - simulating processing lag sleep=8.0 seconds
2017-12-14 00:42:32,962 - jtoc - INFO - hd msg=e72e09da34_1 from_ex=ecomm.api from_rk=ecomm.api.west
2017-12-14 00:42:32,963 - jtoc - INFO - relay msg_id=e72e09da34_1 body={'msg_id': 'e72e09da34_1', 've
2017-12-14 00:42:32,968 - jtoc - INFO - relay done with msg_id=e72e09da34_1
2017-12-14 00:42:32,968 - jtoc - INFO - task - ecomm_app.ecommerce.tasks.handle_user_conversion_events - simulating processing lag sleep=8.0 seconds
2017-12-14 00:42:40,982 - jtoc - INFO - hd msg=f5ec3f0c9d_1 from_ex=ecomm.api from_rk=ecomm.api.west
2017-12-14 00:42:40,983 - jtoc - INFO - relay msg_id=f5ec3f0c9d_1 body={'msg_id': 'f5ec3f0c9d_1', 've
2017-12-14 00:42:41,005 - jtoc - INFO - relay done with msg_id=f5ec3f0c9d_1
2017-12-14 00:42:41,006 - jtoc - INFO - task - ecomm_app.ecommerce.tasks.handle_user_conversion_events - simulating processing lag sleep=8.0 seconds
2017-12-14 00:42:49,014 - jtoc - INFO - hd msg=222094db10_1 from_ex=ecomm.api from_rk=ecomm.api.west
2017-12-14 00:42:49,015 - jtoc - INFO - relay msg_id=222094db10_1 body={'msg_id': '222094db10_1', 've
2017-12-14 00:42:49,024 - jtoc - INFO - relay done with msg_id=222094db10_1
2017-12-14 00:42:49,024 - jtoc - INFO - task - ecomm_app.ecommerce.tasks.handle_user_conversion_events - simulating processing lag sleep=8.0 seconds
2017-12-14 00:42:57,034 - jtoc - INFO - hd msg=9bed4cc0e5_1 from_ex=ecomm.api from_rk=ecomm.api.west
2017-12-14 00:42:57,035 - jtoc - INFO - relay msg_id=9bed4cc0e5_1 body={'msg_id': '9bed4cc0e5_1', 've
2017-12-14 00:42:57,045 - jtoc - INFO - relay done with msg_id=9bed4cc0e5_1
2017-12-14 00:42:57,045 - jtoc - INFO - task - ecomm_app.ecommerce.tasks.handle_user_conversion_events - simulating processing lag sleep=8.0 seconds
2017-12-14 00:43:05,052 - jtoc - INFO - hd msg=f66139a9cf_1 from_ex=ecomm.api from_rk=ecomm.api.west
2017-12-14 00:43:05,053 - jtoc - INFO - relay msg_id=f66139a9cf_1 body={'msg_id': 'f66139a9cf_1', 've
2017-12-14 00:43:05,061 - jtoc - INFO - relay done with msg_id=f66139a9cf_1
2017-12-14 00:43:05,061 - jtoc - INFO - task - ecomm_app.ecommerce.tasks.handle_user_conversion_events - simulating processing lag sleep=8.0 seconds
2017-12-14 00:43:13,073 - jtoc - INFO - hd msg=94d3a2c7ed_1 from_ex=ecomm.api from_rk=ecomm.api.west
2017-12-14 00:43:13,074 - jtoc - INFO - relay msg_id=94d3a2c7ed_1 body={'msg_id': '94d3a2c7ed_1', 've
2017-12-14 00:43:13,095 - jtoc - INFO - relay done with msg_id=94d3a2c7ed_1
2017-12-14 00:43:13,098 - jtoc - INFO - task - ecomm_app.ecommerce.tasks.handle_user_conversion_events - simulating processing lag sleep=8.0 seconds
2017-12-14 00:43:21,105 - jtoc - INFO - hd msg=b517f87ff4_1 from_ex=ecomm.api from_rk=ecomm.api.west
2017-12-14 00:43:21,106 - jtoc - INFO - relay msg_id=b517f87ff4_1 body={'msg_id': 'b517f87ff4_1', 've
2017-12-14 00:43:21,123 - jtoc - INFO - relay done with msg_id=b517f87ff4_1
2017-12-14 00:43:21,124 - jtoc - INFO - task - ecomm_app.ecommerce.tasks.handle_user_conversion_events - simulating processing lag sleep=8.0 seconds
2017-12-14 00:43:29,140 - jtoc - INFO - hd msg=822ef4142c_1 from_ex=ecomm.api from_rk=ecomm.api.west
2017-12-14 00:43:29,140 - jtoc - INFO - relay msg_id=822ef4142c_1 body={'msg_id': '822ef4142c_1', 've
2017-12-14 00:43:29,147 - jtoc - INFO - relay done with msg_id=822ef4142c_1
2017-12-14 00:43:29,147 - jtoc - INFO - task - ecomm_app.ecommerce.tasks.handle_user_conversion_events - simulating processing lag sleep=8.0 seconds

Verify the Ecomm Celery Application Processed the Task

The logs for the ecomm Celery worker should show something similar to:

[2017-12-14 00:42:16,938: INFO/MainProcess] Received task: ecomm_app.ecommerce.tasks.handle_user_conversion_events[7848f13c-00e1-47d1-b5a5-a8e0dea1dc04]   expires:[2017-12-14 08:47:16.881373+00:00]
[2017-12-14 00:42:16,940: INFO/ForkPoolWorker-1] task - user_conversion_events - start body={'subscription_id': 321, 'r_id': '2483467dad_1', 'stripe_id': 876, 'version': 1, 'created': '2017-12-14T00:42:16.870156', 'product_id': 'JJJ', 'account_id': 999, 'msg_id': '46cb24f0d0_1'}
[2017-12-14 00:42:16,940: INFO/ForkPoolWorker-1] task - user_conversion_events - done
[2017-12-14 00:42:16,942: INFO/ForkPoolWorker-1] Task ecomm_app.ecommerce.tasks.handle_user_conversion_events[7848f13c-00e1-47d1-b5a5-a8e0dea1dc04] succeeded in 0.002363318002608139s: True
[2017-12-14 00:42:24,954: INFO/MainProcess] Received task: ecomm_app.ecommerce.tasks.handle_user_conversion_events[4ea2b08f-efa1-46f8-a522-7e2ccde37f4e]   expires:[2017-12-14 08:47:24.950295+00:00]
[2017-12-14 00:42:24,955: INFO/ForkPoolWorker-2] task - user_conversion_events - start body={'subscription_id': 321, 'r_id': '88daa66cac_1', 'stripe_id': 876, 'version': 1, 'created': '2017-12-14T00:42:24.947811', 'product_id': 'JJJ', 'account_id': 999, 'msg_id': 'd2724b75fa_1'}
[2017-12-14 00:42:24,955: INFO/ForkPoolWorker-2] task - user_conversion_events - done
[2017-12-14 00:42:24,960: INFO/ForkPoolWorker-2] Task ecomm_app.ecommerce.tasks.handle_user_conversion_events[4ea2b08f-efa1-46f8-a522-7e2ccde37f4e] succeeded in 0.005053305001638364s: True
[2017-12-14 00:42:32,979: INFO/MainProcess] Received task: ecomm_app.ecommerce.tasks.handle_user_conversion_events[496e6c89-e725-433d-8bfa-a0d0decc8e0d]   expires:[2017-12-14 08:47:32.965396+00:00]
[2017-12-14 00:42:32,981: INFO/ForkPoolWorker-3] task - user_conversion_events - start body={'subscription_id': 321, 'r_id': '2bb5cdd264_1', 'stripe_id': 876, 'version': 1, 'created': '2017-12-14T00:42:32.963186', 'product_id': 'JJJ', 'account_id': 999, 'msg_id': 'e72e09da34_1'}
[2017-12-14 00:42:32,981: INFO/ForkPoolWorker-3] task - user_conversion_events - done
[2017-12-14 00:42:32,987: INFO/ForkPoolWorker-3] Task ecomm_app.ecommerce.tasks.handle_user_conversion_events[496e6c89-e725-433d-8bfa-a0d0decc8e0d] succeeded in 0.00654161800048314s: True
[2017-12-14 00:42:41,008: INFO/MainProcess] Received task: ecomm_app.ecommerce.tasks.handle_user_conversion_events[f4f7681e-1bca-4798-a73a-89f62317651d]   expires:[2017-12-14 08:47:40.991378+00:00]
[2017-12-14 00:42:41,012: INFO/ForkPoolWorker-1] task - user_conversion_events - start body={'subscription_id': 321, 'r_id': '5365dc6b70_1', 'stripe_id': 876, 'version': 1, 'created': '2017-12-14T00:42:40.983174', 'product_id': 'JJJ', 'account_id': 999, 'msg_id': 'f5ec3f0c9d_1'}
[2017-12-14 00:42:41,012: INFO/ForkPoolWorker-1] task - user_conversion_events - done
[2017-12-14 00:42:41,014: INFO/ForkPoolWorker-1] Task ecomm_app.ecommerce.tasks.handle_user_conversion_events[f4f7681e-1bca-4798-a73a-89f62317651d] succeeded in 0.002192696003476158s: True
[2017-12-14 00:42:49,026: INFO/MainProcess] Received task: ecomm_app.ecommerce.tasks.handle_user_conversion_events[35d5ed9e-aacf-4b05-bae0-f74b8df83ad2]   expires:[2017-12-14 08:47:49.017937+00:00]
[2017-12-14 00:42:49,028: INFO/ForkPoolWorker-2] task - user_conversion_events - start body={'subscription_id': 321, 'r_id': 'f369b4c0e0_1', 'stripe_id': 876, 'version': 1, 'created': '2017-12-14T00:42:49.015218', 'product_id': 'JJJ', 'account_id': 999, 'msg_id': '222094db10_1'}
[2017-12-14 00:42:49,028: INFO/ForkPoolWorker-2] task - user_conversion_events - done
[2017-12-14 00:42:49,031: INFO/ForkPoolWorker-2] Task ecomm_app.ecommerce.tasks.handle_user_conversion_events[35d5ed9e-aacf-4b05-bae0-f74b8df83ad2] succeeded in 0.00297039799625054s: True
[2017-12-14 00:42:57,047: INFO/MainProcess] Received task: ecomm_app.ecommerce.tasks.handle_user_conversion_events[4c9137a1-e4c0-44f2-852d-b96e8004cf52]   expires:[2017-12-14 08:47:57.040272+00:00]
[2017-12-14 00:42:57,050: INFO/ForkPoolWorker-3] task - user_conversion_events - start body={'subscription_id': 321, 'r_id': '81646a1d3e_1', 'stripe_id': 876, 'version': 1, 'created': '2017-12-14T00:42:57.035385', 'product_id': 'JJJ', 'account_id': 999, 'msg_id': '9bed4cc0e5_1'}
[2017-12-14 00:42:57,051: INFO/ForkPoolWorker-3] task - user_conversion_events - done
[2017-12-14 00:42:57,053: INFO/ForkPoolWorker-3] Task ecomm_app.ecommerce.tasks.handle_user_conversion_events[4c9137a1-e4c0-44f2-852d-b96e8004cf52] succeeded in 0.0024162650006473996s: True
[2017-12-14 00:43:05,061: INFO/MainProcess] Received task: ecomm_app.ecommerce.tasks.handle_user_conversion_events[c0eafd5b-803a-4550-9bba-961d7ab7f4cc]   expires:[2017-12-14 08:48:05.056204+00:00]
[2017-12-14 00:43:05,064: INFO/ForkPoolWorker-1] task - user_conversion_events - start body={'subscription_id': 321, 'r_id': '360be5bb5d_1', 'stripe_id': 876, 'version': 1, 'created': '2017-12-14T00:43:05.052968', 'product_id': 'JJJ', 'account_id': 999, 'msg_id': 'f66139a9cf_1'}
[2017-12-14 00:43:05,065: INFO/ForkPoolWorker-1] task - user_conversion_events - done
[2017-12-14 00:43:05,067: INFO/ForkPoolWorker-1] Task ecomm_app.ecommerce.tasks.handle_user_conversion_events[c0eafd5b-803a-4550-9bba-961d7ab7f4cc] succeeded in 0.003034861001651734s: True
[2017-12-14 00:43:13,100: INFO/MainProcess] Received task: ecomm_app.ecommerce.tasks.handle_user_conversion_events[b402c99b-b998-48b8-9cb8-bb49b1289032]   expires:[2017-12-14 08:48:13.081228+00:00]
[2017-12-14 00:43:13,106: INFO/ForkPoolWorker-2] task - user_conversion_events - start body={'subscription_id': 321, 'r_id': '5d4d3f1277_1', 'stripe_id': 876, 'version': 1, 'created': '2017-12-14T00:43:13.074799', 'product_id': 'JJJ', 'account_id': 999, 'msg_id': '94d3a2c7ed_1'}
[2017-12-14 00:43:13,107: INFO/ForkPoolWorker-2] task - user_conversion_events - done
[2017-12-14 00:43:13,110: INFO/ForkPoolWorker-2] Task ecomm_app.ecommerce.tasks.handle_user_conversion_events[b402c99b-b998-48b8-9cb8-bb49b1289032] succeeded in 0.004359455000667367s: True
[2017-12-14 00:43:21,127: INFO/MainProcess] Received task: ecomm_app.ecommerce.tasks.handle_user_conversion_events[a57b8b49-349a-44d3-99fc-12f96d69d489]   expires:[2017-12-14 08:48:21.114216+00:00]
[2017-12-14 00:43:21,129: INFO/ForkPoolWorker-3] task - user_conversion_events - start body={'subscription_id': 321, 'r_id': '97ec19ac8d_1', 'stripe_id': 876, 'version': 1, 'created': '2017-12-14T00:43:21.106783', 'product_id': 'JJJ', 'account_id': 999, 'msg_id': 'b517f87ff4_1'}
[2017-12-14 00:43:21,130: INFO/ForkPoolWorker-3] task - user_conversion_events - done
[2017-12-14 00:43:21,133: INFO/ForkPoolWorker-3] Task ecomm_app.ecommerce.tasks.handle_user_conversion_events[a57b8b49-349a-44d3-99fc-12f96d69d489] succeeded in 0.003475217003142461s: True
[2017-12-14 00:43:29,150: INFO/MainProcess] Received task: ecomm_app.ecommerce.tasks.handle_user_conversion_events[da0c3a78-cac7-4b78-8a32-568d8a5c7362]   expires:[2017-12-14 08:48:29.143188+00:00]
[2017-12-14 00:43:29,152: INFO/ForkPoolWorker-1] task - user_conversion_events - start body={'subscription_id': 321, 'r_id': '6f3fe96baf_1', 'stripe_id': 876, 'version': 1, 'created': '2017-12-14T00:43:29.140647', 'product_id': 'JJJ', 'account_id': 999, 'msg_id': '822ef4142c_1'}
[2017-12-14 00:43:29,152: INFO/ForkPoolWorker-1] task - user_conversion_events - done
[2017-12-14 00:43:29,155: INFO/ForkPoolWorker-1] Task ecomm_app.ecommerce.tasks.handle_user_conversion_events[da0c3a78-cac7-4b78-8a32-568d8a5c7362] succeeded in 0.0034106169987353496s: True

Benchmark the JSON to Celery Relay Service

The start-mixin-load-test.py load test will send in 20,000 messages with no simulated lag. This may take a few moments to finish so you might want to open a new terminal and source the virtual env to run watch -n5 list-queues.sh for tracking the test’s progress.

./start-mixin-load-test.py
2017-12-14 00:48:06,217 - robopub - INFO - Generating messages=20000
2017-12-14 00:48:06,694 - robopub - INFO - Publishing messages=20000
2017-12-14 00:48:06,821 - pub - INFO - 1.00 send done msg=200/20000 ex=ecomm.api rk=ecomm.api.west
2017-12-14 00:48:06,821 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=69ae9e80bf_1
2017-12-14 00:48:06,916 - pub - INFO - 2.00 send done msg=400/20000 ex=ecomm.api rk=ecomm.api.west
2017-12-14 00:48:06,917 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=43c153a155_1
2017-12-14 00:48:07,015 - pub - INFO - 3.00 send done msg=600/20000 ex=ecomm.api rk=ecomm.api.west
2017-12-14 00:48:07,016 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=1978ebf438_1
2017-12-14 00:48:07,075 - pub - INFO - 4.00 send done msg=800/20000 ex=ecomm.api rk=ecomm.api.west
2017-12-14 00:48:07,075 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=3dbae69bb2_1
2017-12-14 00:48:07,157 - pub - INFO - 5.00 send done msg=1000/20000 ex=ecomm.api rk=ecomm.api.west
2017-12-14 00:48:07,158 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=a5dda8b23a_1
2017-12-14 00:48:07,240 - pub - INFO - 6.00 send done msg=1200/20000 ex=ecomm.api rk=ecomm.api.west
2017-12-14 00:48:07,241 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=138a5c7939_1
2017-12-14 00:48:07,310 - pub - INFO - 7.00 send done msg=1400/20000 ex=ecomm.api rk=ecomm.api.west
2017-12-14 00:48:07,311 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=a1c6315380_1
2017-12-14 00:48:07,374 - pub - INFO - 8.00 send done msg=1600/20000 ex=ecomm.api rk=ecomm.api.west
2017-12-14 00:48:07,374 - pub - INFO - ex=ecomm.api rk=ecomm.api.west msg=f1cf343847_1

Or with docker compose:

docker-compose -f compose-start-mixin-load-test.yml up
WARNING: Found orphan containers (ecommworker, jtocrelay) for this project. If you removed or renamed this service in your compose file, you can run this command with the --remove-orphans flag to clean it up.
Starting mixinloadtest ... done
Attaching to mixinloadtest

Sample output during that load test:

list-queues.sh

Listing Queues broker=localhost:15672
name durable auto_delete consumers messages messages_ready messages_unacknowledged
celery True False 1 0 0 0
celery@ecommerce_subscriber.celery.pidbox False True 1 0 0 0
celeryev.28b0b3a0-2e82-4e16-b829-a2835763b3cb False True 1 0 0 0
celeryev.b019122d-0dd3-48c0-8c0a-b82f4fb8d4d7 False True 1 0 0 0
ecomm.api.west True False 1 17810 17809 1

View the Ecomm Celery Worker Tasks in Flower

You can also watch progress using the Flower Celery monitoring application that’s included in the docker compose file.

Here’s a snapshot of my 20,000 + 10 messages using the celery@ecommerce_subscriber Celery worker.

https://raw.githubusercontent.com/jay-johnson/celery-connectors/master/_images/flower-jtoc-relay-results.png

The Processed and Succeeded task counts for the celery@ecommerce_subscriber should increment each time a User Conversion Event is published by the ecomm relay to the ecomm worker.

http://localhost:5555/dashboard (login: admin/admin)

View specific task details:

http://localhost:5555/tasks

Stop the Ecomm Demo

Restart the docker containers to a good, clean state.

Stop:

stop-redis-and-rabbitmq.sh
Stopping redis and rabbitmq
Stopping celrabbit1      ... done
Stopping celredis1       ... done
Stopping celflowerredis  ... done
Stopping celflowerrabbit ... done

Start:

start-redis-and-rabbitmq.sh
Starting redis and rabbitmq
Creating celrabbit1 ... done

Verify the Relay Service Automatically Healed

Want to try the load test again now that we just simulated a broker outage for all of the messaging and monitoring containers?

./start-mixin-load-test.py

or

docker-compose -f compose-start-mixin-load-test.yml up

If not, then stop the ecomm relay and ecomm worker terminal sessions using: ctrl + c

Running an Ecommerce JSON-to-Celery Relay Service - Example 2

This example uses just kombu producers and consumers instead of the kombu.ConsumerProducerMixin to run the same relay as the example above.

Start Ecommerce Celery Worker

Start a Celery worker for an existing ecommerce application from a hypothetical Django or Flask server.

Note: Please run this from the base directory for the repository and source the virtual env: source venv/bin/activate

./start-ecomm-worker.sh

-------------- celery@ecommerce_subscriber v4.1.0 (latentcall)
---- **** -----
--- * ***  * -- Linux-4.7.4-200.fc24.x86_64-x86_64-with-fedora-24-Twenty_Four 2017-12-14 00:33:02
-- * - **** ---
- ** ---------- [config]
- ** ---------- .> app:         ecommerce-worker:0x7f0c23f1c550
- ** ---------- .> transport:   amqp://rabbitmq:**@localhost:5672//
- ** ---------- .> results:     redis://localhost:6379/10
- *** --- * --- .> concurrency: 3 (prefork)
-- ******* ---- .> task events: OFF (enable -E to monitor tasks in this worker)
--- ***** -----
-------------- [queues]
                .> celery           exchange=celery(direct) key=celery


[tasks]
. ecomm_app.ecommerce.tasks.handle_user_conversion_events

[2017-12-14 00:33:02,243: INFO/MainProcess] Connected to amqp://rabbitmq:**@127.0.0.1:5672//
[2017-12-14 00:33:02,260: INFO/MainProcess] mingle: searching for neighbors
[2017-12-14 00:33:03,293: INFO/MainProcess] mingle: all alone
[2017-12-14 00:33:03,337: INFO/MainProcess] celery@ecommerce_subscriber ready.
[2017-12-14 00:33:05,275: INFO/MainProcess] Events of group {task} enabled by remote.

Or with docker compose:

docker-compose -f compose-start-ecomm-worker.yml up
Starting ecommworker ... done
Attaching to ecommworker

Notice the worker is named celery@ecommerce_subscriber this is the identifier for viewing the Celery application in Flower:

http://localhost:5555/worker/celery@ecommerce_subscriber (login: admin/admin)

Start Ecomm Relay

This process will consume JSON dictionary messages on the user.events.conversions RabbitMQ queue and pass the message to the ecomm Celery app as a ecomm_app.ecommerce.tasks.handle_user_conversion_events Celery task.

Please start this in a new terminal that has sourced the virtual env: source venv/bin/activate

./start-ecomm-relay.py
2017-12-14 00:33:36,943 - ecomm-relay-loader - INFO - Start - ecomm-relay
2017-12-14 00:33:36,944 - message-processor - INFO - ecomm-relay START - consume_queue=user.events.conversions rk=reporting.accounts callback=relay_callback
2017-12-14 00:33:36,944 - kombu-subscriber - INFO - setup routing

Or with docker compose:

docker-compose -f compose-start-ecomm-relay.yml up
WARNING: Found orphan containers (ecommworker) for this project. If you removed or renamed this service in your compose file, you can run this command with the --remove-orphans flag to clean it up.
Creating ecommrelay ... done
Attaching to ecommrelay

Publish a User Conversion Event to the Ecomm Relay

This will use Kombu to publish a JSON dictionary message to the user.events.conversions RabbitMQ queue which is monitored by the ecomm relay.

Please start this in a new terminal that has sourced the virtual env: source venv/bin/activate

publish-user-conversion-events-rabbitmq.py
INFO:publish-user-conversion-events:Start - publish-user-conversion-events
INFO:publish-user-conversion-events:Sending user conversion event msg={'product_id': 'XYZ', 'stripe_id': 999, 'account_id': 777, 'created': '2017-12-14T00:33:55.826534', 'subscription_id': 888} ex=user.events rk=user.events.conversions
INFO:kombu-publisher:SEND - exch=user.events rk=user.events.conversions
INFO:publish-user-conversion-events:End - publish-user-conversion-events sent=True

Or with docker compose:

docker-compose -f compose-publish-user-conversion-events-rabbitmq.yml up
WARNING: Found orphan containers (ecommrelay, ecommworker) for this project. If you removed or renamed this service in your compose file, you can run this command with the --remove-orphans flag to clean it up.
Starting ucepubrmq ... done
Attaching to ucepubrmq

Verify the Ecomm Relay Processed the Conversion Message

The logs for the ecomm relay should show something similar to:

2017-12-14 00:33:55,865 - ecomm-relay-loader - INFO - Sending broker=amqp://rabbitmq:rabbitmq@localhost:5672// body={'org_msg': {'stripe_id': 999, 'created': '2017-12-14T00:33:55.826534', 'product_id': 'XYZ', 'subscription_id': 888, 'account_id': 777}, 'stripe_id': 876, 'version': 1, 'account_id': 999, 'msg_id': '7a73a74d-f539-4634-8a03-2aa2a5fd8d5e', 'created': '2017-12-14T00:33:55.863870', 'product_id': 'JJJ', 'subscription_id': 321}
2017-12-14 00:33:55,928 - ecomm-relay-loader - INFO - Done with msg_id=7a73a74d-f539-4634-8a03-2aa2a5fd8d5e result=True

Verify the Ecomm Celery Application Processed the Task

The logs for the ecomm Celery worker should show something similar to:

[2017-12-14 00:33:55,919: INFO/MainProcess] Received task: ecomm_app.ecommerce.tasks.handle_user_conversion_events[9ee85235-0ffb-4c46-9cd7-0bd2c153bd9b]
[2017-12-14 00:33:55,921: INFO/ForkPoolWorker-1] task - user_conversion_events - start body={'subscription_id': 321, 'stripe_id': 876, 'org_msg': {'subscription_id': 888, 'stripe_id': 999, 'created': '2017-12-14T00:33:55.826534', 'product_id': 'XYZ', 'account_id': 777}, 'version': 1, 'created': '2017-12-14T00:33:55.863870', 'product_id': 'JJJ', 'account_id': 999, 'msg_id': '7a73a74d-f539-4634-8a03-2aa2a5fd8d5e'}
[2017-12-14 00:33:55,921: INFO/ForkPoolWorker-1] task - user_conversion_events - done
[2017-12-14 00:33:55,926: INFO/ForkPoolWorker-1] Task ecomm_app.ecommerce.tasks.handle_user_conversion_events[9ee85235-0ffb-4c46-9cd7-0bd2c153bd9b] succeeded in 0.0055257950007217005s: True

View the Ecomm Celery Worker Tasks in Flower

The Processed and Succeeded task counts for the celery@ecommerce_subscriber should increment each time a User Conversion Event is published by the ecomm relay to the ecomm worker.

http://localhost:5555/dashboard

View specific task details:

http://localhost:5555/tasks

Stop the Ecomm Demo

In the ecomm relay and ecomm worker terminal sessions use: ctrl + c to stop the processes.

Restart the docker containers to a good, clean state.

Stop:

stop-redis-and-rabbitmq.sh
Stopping redis and rabbitmq
Stopping celrabbit1      ... done
Stopping celredis1       ... done
Stopping celflowerredis  ... done
Stopping celflowerrabbit ... done

Start:

start-redis-and-rabbitmq.sh
Starting redis and rabbitmq
Creating celrabbit1 ... done

View the Ecomm Celery Worker Tasks in Flower

The Processed and Succeeded task counts for the celery@ecommerce_subscriber should increment each time a User Conversion Event is published by the ecomm relay to the ecomm worker.

http://localhost:5555/dashboard

View specific task details:

http://localhost:5555/tasks

Stop the Ecomm Demo

In the ecomm relay and ecomm worker terminal sessions use: ctrl + c to stop the processes.

Restart the docker containers to a good, clean state.

Stop:

stop-redis-and-rabbitmq.sh
Stopping redis and rabbitmq
Stopping celrabbit1      ... done
Stopping celredis1       ... done
Stopping celflowerredis  ... done
Stopping celflowerrabbit ... done

Start:

start-redis-and-rabbitmq.sh
Starting redis and rabbitmq
Creating celrabbit1 ... done
Creating celrabbit1 ... done
Creating celredis1 ... done
Creating celflowerredis ... done

Celery Bootstep with RabbitMQ Outage Example

This example uses Celery bootsteps (http://docs.celeryproject.org/en/latest/userguide/extending.html) to run a standalone, headless subscriber that consumes routed messages to two queues. It will set up a RabbitMQ topic exchange with a queue that is bound using a routing key and a separate direct queue for additional messages to process. Once the entities are available in RabbitMQ, Kombu publishes the message to the exchanges and RabbitMQ provides the messaging facility to route the messages to the subscribed Celery workers’ queues. Once messages are being processed we will simulate a broker failure and see how resilient Celery bootsteps are to this type of disaster.

  1. Stop and Start the docker containers

    ./stop-redis-and-rabbitmq.sh
    Stopping redis and rabbitmq
    Stopping celredis1       ... done
    Stopping celflowerrabbit ... done
    Stopping celflowerredis  ... done
    Stopping celrabbit1      ... done
    
    ./start-redis-and-rabbitmq.sh
    Starting redis and rabbitmq
    Creating celrabbit1 ... done
    Creating celredis1 ...
    Creating celflowerredis ...
    Creating celrabbit1 ...
    
  2. List the Queues

    list-queues.sh
    

    Listing Queues broker=localhost:15672

    name durable auto_delete consumers messages messages_ready messages_unacknowledged
    celeryev.a1ccb5f7-4f76-4e26-9cdc-bf5438ba5362 False True 1 0 0 0
  3. Publish a message

    run_rabbitmq_publisher.py
    INFO:run-rabbitmq-publisher:Start - run-rabbitmq-publisher
    INFO:run-rabbitmq-publisher:Sending msg={'created': '2017-12-14T18:08:29.481313', 'account_id': 456} ex=reporting rk=reporting.accounts
    INFO:kombu-publisher:SEND - exch=reporting rk=reporting.accounts
    INFO:run-rabbitmq-publisher:End - run-rabbitmq-publisher sent=True
    

    Or with docker compose:

    docker-compose -f compose-run-rabbitmq-publisher.yml up
    Creating kombupubrmq ... done
    Attaching to kombupubrmq
    kombupubrmq    | 2017-12-15 07:31:23,802 - run-rabbitmq-publisher - INFO - Start - run-rabbitmq-publisher
    kombupubrmq    | 2017-12-15 07:31:23,802 - run-rabbitmq-publisher - INFO - Sending msg={'account_id': 456, 'created': '2017-12-15T07:31:23.802616'} ex=reporting rk=reporting.accounts
    kombupubrmq    | 2017-12-15 07:31:23,899 - kombu-publisher - INFO - SEND - exch=reporting rk=reporting.accounts
    kombupubrmq    | 2017-12-15 07:31:23,903 - run-rabbitmq-publisher - INFO - End - run-rabbitmq-publisher sent=True
    kombupubrmq exited with code 0
    
  4. Confirm the message is ready in the RabbitMQ Queue

    Note the messages and messages_ready count increased while the messages_unacknowledged did not. Which is because we have not started the subscriber to process ready messages in the reporting.accounts queue.

    list-queues.sh
    

    Listing Queues broker=localhost:15672

    name durable auto_delete consumers messages messages_ready messages_unacknowledged
    celeryev.a1ccb5f7-4f76-4e26-9cdc-bf5438ba5362 False True 1 0 0 0
    reporting.accounts True False 0 1 1 0
  5. List the Exchanges

    list-exchanges.sh
    

    Listing Exchanges broker=localhost:15672

    name type durable auto_delete
      direct True False
    amq.direct direct True False
    amq.fanout fanout True False
    amq.headers headers True False
    amq.match headers True False
    amq.rabbitmq.log topic True False
    amq.rabbitmq.trace topic True False
    amq.topic topic True False
    celery.pidbox fanout False False
    celeryev topic True False
    reply.celery.pidbox direct False False
    reporting topic True False
  6. Consume that message by starting up the Celery Rabbitmq subscriber module

    This will consume messages from the reporting.accounts and reporting.subscriptions queues.

    celery worker -A run_rabbitmq_subscriber -n rabbitmq_bootstep -c 3 --loglevel=INFO -Ofair
    

    Or with docker compose:

    docker-compose -f compose-run-celery-rabbitmq-subscriber.yml up
    Creating celeryrabbitmqsubscriber ... done
    Attaching to celeryrabbitmqsubscriber
    
  7. Confirm the worker’s logs show the message was received

    2017-12-14 10:10:25,832: INFO callback received msg body={'account_id': 456, 'created': '2017-12-14T18:08:29.481313'} from_ex=reporting from_rk=reporting.accounts
    
  8. View the Rabbit Subscriber celery@rabbitmq_bootstep in Flower

    Rabbit Flower server (login admin/admin)

    http://localhost:5555/

  9. Verify the message is no longer in the Queue and Celery is connected as a consumer

    With the Celery RabbitMQ worker still running, in a new terminal list the queues. Verify there is a consumer on the reporting.accounts and reporting.subscriptions queues.

    list-queues.sh
    

    Listing Queues broker=localhost:15672

    name durable auto_delete consumers messages messages_ready messages_unacknowledged
    celery.rabbit.sub True False 1 0 0 0
    celery@rabbitmq_bootstep.celery.pidbox False True 1 0 0 0
    celeryev.a1ccb5f7-4f76-4e26-9cdc-bf5438ba5362 False True 1 0 0 0
    celeryev.f85fe29a-b729-48fa-a17d-b7e12c14dba8 False True 1 0 0 0
    reporting.accounts True False 1 0 0 0
    reporting.subscriptions True False 1 0 0 0
  10. Start the Queue watcher

    In a new terminal activate the virtual env source venv/bin/activate.

    watch-queues.sh
    

    The watch will poll RabbitMQ for the queues every second and before the load tests start should look empty:

    name durable auto_delete consumers messages messages_ready messages_unacknowledged
    celery.rabbit.sub True False 1 0 0 0
    celery@rabbitmq_bootstep.celery.pidbox False True 1 0 0 0
    celeryev.a1ccb5f7-4f76-4e26-9cdc-bf5438ba5362 False True 1 0 0 0
    celeryev.f85fe29a-b729-48fa-a17d-b7e12c14dba8 False True 1 0 0 0
    reporting.accounts True False 1 0 0 0
    reporting.subscriptions True False 1 0 0 0
  11. Start the Accounts and Subscriptions Load Tests

    This will require two separate terminal sessions with the virtual env activated source venv/bin/activate.

    In terminal 1 start the Accounts load test

    start-load-test-rabbitmq.py
    

    Or with docker compose:

    docker-compose -f compose-start-load-test-rabbitmq.yml up
    WARNING: Found orphan containers (subsloadtest, celeryrabbitmqsubscriber) for this project. If you removed or renamed this service in your compose file, you can run this command with the --remove-orphans flag to clean it up.
    Creating loadtestrmq ... done
    Attaching to loadtestrmq
    

    In terminal 2 start the Subscriptions load test

    start-subscriptions-rabbitmq-test.py
    

    Or with docker compose:

    docker-compose -f compose-start-subscriptions-rabbitmq-test.yml up
    WARNING: Found orphan containers (celeryrabbitmqsubscriber) for this project. If you removed or renamed this service in your compose file, you can run this command with the --remove-orphans flag to clean it up.
    Creating subsloadtest ... done
    Attaching to subsloadtest
    
  12. Verify the Queues are filling up

    After a few seconds, the queues should be filling up with Account and Subscription messages that are being actively processed.

    name durable auto_delete consumers messages messages_ready messages_unacknowledged
    celery.rabbit.sub True False 1 0 0 0
    celery@rabbitmq_bootstep.celery.pidbox False True 1 0 0 0
    celeryev.a1ccb5f7-4f76-4e26-9cdc-bf5438ba5362 False True 1 0 0 0
    celeryev.f85fe29a-b729-48fa-a17d-b7e12c14dba8 False True 1 0 0 0
    reporting.accounts True False 1 31157 31154 3
    reporting.subscriptions True False 1 30280 30277 3
  13. Verify the Celery Bootstep Subscriber is processing messages

    By default the Celery subscriber workers are processing 1 message at a time per consumer. In this example we started 3 workers so there are 3 messages that are unacknowledged at a time. Confirm messages are being processed from_rk=reporting.subscriptions and from_rk=reporting.accounts. This means the Celery workers are processing messages that have routing keys from the different queues.

    2017-12-14 10:24:12,168: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.178164', 'msg_id': '66e0d69aa0_1'} from_ex= from_rk=reporting.subscriptions
    2017-12-14 10:24:12,168: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.845445', 'msg_id': 'd64e18e7be_1'} from_ex= from_rk=reporting.accounts
    2017-12-14 10:24:12,169: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.178278', 'msg_id': '712132669a_1'} from_ex= from_rk=reporting.subscriptions
    2017-12-14 10:24:12,170: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.845478', 'msg_id': '2174427099_1'} from_ex= from_rk=reporting.accounts
    2017-12-14 10:24:12,182: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.178345', 'msg_id': '1d4a251145_1'} from_ex= from_rk=reporting.subscriptions
    2017-12-14 10:24:12,183: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.178380', 'msg_id': 'b62922b333_1'} from_ex= from_rk=reporting.subscriptions
    2017-12-14 10:24:12,184: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.178404', 'msg_id': 'adc1b1988e_1'} from_ex= from_rk=reporting.subscriptions
    2017-12-14 10:24:12,184: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.845491', 'msg_id': '91ac6c413c_1'} from_ex= from_rk=reporting.accounts
    2017-12-14 10:24:12,184: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.845505', 'msg_id': '0ffd4abf90_1'} from_ex= from_rk=reporting.accounts
    2017-12-14 10:24:12,185: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.845519', 'msg_id': '5a11d2aa97_1'} from_ex= from_rk=reporting.accounts
    2017-12-14 10:24:12,185: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.178843', 'msg_id': '77dd35ade4_1'} from_ex= from_rk=reporting.subscriptions
    2017-12-14 10:24:12,186: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.178944', 'msg_id': '2317ff179d_1'} from_ex= from_rk=reporting.subscriptions
    2017-12-14 10:24:12,186: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.179021', 'msg_id': 'acce2d2672_1'} from_ex= from_rk=reporting.subscriptions
    
  14. Stop the Docker containers

    Note: you can stop the docker containers while the tests are still publishing messages if you want. They should gracefully reconnect once the broker is restored.

    ./stop-redis-and-rabbitmq.sh
    Stopping redis and rabbitmq
    Stopping celflowerredis  ... done
    Stopping celflowerrabbit ... done
    Stopping celrabbit1      ... done
    Stopping celredis1       ... done
    
  15. Confirm Celery was disconnected

    2017-12-14 10:27:00,213: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.995821', 'msg_id': 'a138cf8d8c_1'} from_ex= from_rk=reporting.accounts
    2017-12-14 10:27:00,213: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.461333', 'msg_id': '406df22df7_1'} from_ex= from_rk=reporting.subscriptions
    2017-12-14 10:27:00,214: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.461346', 'msg_id': 'a473232ee4_1'} from_ex= from_rk=reporting.subscriptions
    2017-12-14 10:27:00,214: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:21:46.461361', 'msg_id': '12219ca1fd_1'} from_ex= from_rk=reporting.subscriptions
    2017-12-14 10:27:00,330: WARNING consumer: Connection to broker lost. Trying to re-establish the connection...
    Traceback (most recent call last):
    File "/home/driver/dev/celery-connectors/venv/lib/python3.5/site-packages/celery/worker/consumer/consumer.py", line 320, in start
        blueprint.start(self)
    File "/home/driver/dev/celery-connectors/venv/lib/python3.5/site-packages/celery/bootsteps.py", line 119, in start
        step.start(parent)
    File "/home/driver/dev/celery-connectors/venv/lib/python3.5/site-packages/celery/worker/consumer/consumer.py", line 596, in start
        c.loop(*c.loop_args())
    File "/home/driver/dev/celery-connectors/venv/lib/python3.5/site-packages/celery/worker/loops.py", line 88, in asynloop
        next(loop)
    File "/home/driver/dev/celery-connectors/venv/lib/python3.5/site-packages/kombu-4.1.0-py3.5.egg/kombu/async/hub.py", line 354, in create_loop
        cb(*cbargs)
    File "/home/driver/dev/celery-connectors/venv/lib/python3.5/site-packages/kombu-4.1.0-py3.5.egg/kombu/transport/base.py", line 236, in on_readable
        reader(loop)
    File "/home/driver/dev/celery-connectors/venv/lib/python3.5/site-packages/kombu-4.1.0-py3.5.egg/kombu/transport/base.py", line 218, in _read
        drain_events(timeout=0)
    File "/home/driver/dev/celery-connectors/venv/lib/python3.5/site-packages/amqp-2.2.2-py3.5.egg/amqp/connection.py", line 471, in drain_events
        while not self.blocking_read(timeout):
    File "/home/driver/dev/celery-connectors/venv/lib/python3.5/site-packages/amqp-2.2.2-py3.5.egg/amqp/connection.py", line 476, in blocking_read
        frame = self.transport.read_frame()
    File "/home/driver/dev/celery-connectors/venv/lib/python3.5/site-packages/amqp-2.2.2-py3.5.egg/amqp/transport.py", line 226, in read_frame
        frame_header = read(7, True)
    File "/home/driver/dev/celery-connectors/venv/lib/python3.5/site-packages/amqp-2.2.2-py3.5.egg/amqp/transport.py", line 409, in _read
        raise IOError('Socket closed')
    OSError: Socket closed
    2017-12-14 10:27:00,341: ERROR consumer: Cannot connect to amqp://rabbitmq:**@127.0.0.1:5672//: [Errno 104] Connection reset by peer.
    Trying again in 2.00 seconds...
    
    2017-12-14 10:27:02,369: ERROR consumer: Cannot connect to amqp://rabbitmq:**@127.0.0.1:5672//: [Errno 111] Connection refused.
    Trying again in 4.00 seconds...
    
  16. Start the Docker containers

    ./start-redis-and-rabbitmq.sh
    Starting redis and rabbitmq
    Creating celrabbit1 ... done
    Creating celredis1 ...
    Creating celflowerrabbit ...
    Creating celflowerredis ...
    
  17. Verify the Celery workers reconnected

    2017-12-14 10:28:50,841: INFO Connected to amqp://rabbitmq:**@127.0.0.1:5672//
    2017-12-14 10:28:50,872: INFO mingle: searching for neighbors
    2017-12-14 10:28:51,925: INFO mingle: all alone
    
  18. Start the multi-queue load test publishers again

    In terminal 1:

    ./start-subscriptions-rabbitmq-test.py
    

    Or with docker compose:

    docker-compose -f compose-start-load-test-rabbitmq.yml up
    

    In terminal 2:

    ./start-load-test-rabbitmq.py
    

    Or with docker compose:

    docker-compose -f compose-start-subscriptions-rabbitmq-test.yml up
    
  19. Verify Celery is processing messages from both queues again

    2017-12-14 10:32:19,325: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:31:07.315190', 'msg_id': '22ede22ba6_1'} from_ex= from_rk=reporting.subscriptions
    2017-12-14 10:32:19,326: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:31:07.315213', 'msg_id': '26f1103534_1'} from_ex= from_rk=reporting.subscriptions
    2017-12-14 10:32:19,329: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:31:05.232153', 'msg_id': '10d7a731ca_1'} from_ex= from_rk=reporting.accounts
    2017-12-14 10:32:19,333: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:31:05.232174', 'msg_id': 'ae75ede630_1'} from_ex= from_rk=reporting.accounts
    2017-12-14 10:32:19,336: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:31:07.315225', 'msg_id': '0e86894ae3_1'} from_ex= from_rk=reporting.subscriptions
    2017-12-14 10:32:19,337: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:31:05.232186', 'msg_id': '2066f80569_1'} from_ex= from_rk=reporting.accounts
    2017-12-14 10:32:19,337: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:31:07.315240', 'msg_id': 'ea82241224_1'} from_ex= from_rk=reporting.subscriptions
    2017-12-14 10:32:19,337: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:31:07.315264', 'msg_id': 'accbebead8_1'} from_ex= from_rk=reporting.subscriptions
    2017-12-14 10:32:19,339: INFO callback received msg body={'data': {}, 'created': '2017-12-14T18:31:05.232198', 'msg_id': '8788b7fa97_1'} from_ex= from_rk=reporting.accounts
    

Stop the Celery Bootstep example

In all example terminal sessions, use: ctrl + c to stop any processes you no longer want to run.

Restart the docker containers to a good, clean state for the next example.

Stop:

stop-redis-and-rabbitmq.sh
Stopping redis and rabbitmq
Stopping celrabbit1      ... done
Stopping celredis1       ... done
Stopping celflowerredis  ... done
Stopping celflowerrabbit ... done

Start:

start-redis-and-rabbitmq.sh
Starting redis and rabbitmq
Creating celrabbit1 ... done

Redis Message Processing Example

This example uses Celery bootsteps (http://docs.celeryproject.org/en/latest/userguide/extending.html) to run a standalone, headless subscriber that consumes messages from a Redis key which emulates a RabbitMQ queue. Kombu publishes the message to the Redis key.

  1. Check that the Redis has no keys

    redis-cli
    127.0.0.1:6379> keys *
    (empty list or set)
    127.0.0.1:6379>
    
  2. Publish a message

    run_redis_publisher.py
    2017-12-09 08:20:04,026 - run-redis-publisher - INFO - Start - run-redis-publisher
    2017-12-09 08:20:04,027 - run-redis-publisher - INFO - Sending msg={'account_id': 123, 'created': '2017-12-09T08:20:04.027159'} ex=reporting.accounts rk=reporting.accounts
    2017-12-09 08:20:04,050 - redis-publisher - INFO - SEND - exch=reporting.accounts rk=reporting.accounts
    2017-12-09 08:20:04,052 - run-redis-publisher - INFO - End - run-redis-publisher sent=True
    

    Or with docker compose:

    docker-compose -f compose-run-redis-publisher.yml up
    Creating kombupubredis ... done
    Attaching to kombupubredis
    kombupubredis    | 2017-12-15 07:44:47,047 - run-redis-publisher - INFO - Start - run-redis-publisher
    kombupubredis    | 2017-12-15 07:44:47,047 - run-redis-publisher - INFO - Sending msg={'account_id': 123, 'created': '2017-12-15T07:44:47.047355'} ex=reporting.accounts rk=reporting.accounts
    kombupubredis    | 2017-12-15 07:44:47,127 - kombu-publisher - INFO - SEND - exch=reporting.accounts rk=reporting.accounts
    kombupubredis    | 2017-12-15 07:44:47,132 - run-redis-publisher - INFO - End - run-redis-publisher sent=True
    kombupubredis exited with code 0
    
  3. Consume messages using the subscriber module

    celery worker -A run_redis_subscriber --loglevel=INFO -Ofair
    

    Or with docker compose:

    docker-compose -f compose-run-celery-redis-subscriber.yml up
    WARNING: Found orphan containers (kombupubredis) for this project. If you removed or renamed this service in your compose file, you can run this command with the --remove-orphans flag to clean it up.
    Creating celeryredissubscriber ... done
    Attaching to celeryredissubscriber
    
  4. Confirm the Celery worker received the message

    2017-12-09 08:20:08,221: INFO callback received msg body={u'account_id': 123, u'created': u'2017-12-09T08:20:04.027159'}
    
  5. View the Redis Subscriber in Flower

    Redis Flower server (login admin/admin)

    http://localhost:5556/

  6. Look at the Redis keys

    redis-cli
    127.0.0.1:6379> keys *
    1) "_kombu.binding.celeryev"
    2) "_kombu.binding.celery"
    3) "_kombu.binding.celery.pidbox"
    4) "_kombu.binding.reporting.accounts"
    5) "unacked_mutex"
    127.0.0.1:6379>
    

Redis Kombu Subscriber

If you do not want to use Celery, you can use the KombuSubscriber class to process messages. This class will wait for a configurable amount of seconds to consume a single message from the subscribed queue and then stop processing.

  1. Check the Redis keys

    redis-cli
    127.0.0.1:6379> keys *
    1) "_kombu.binding.reporting.accounts"
    2) "_kombu.binding.celery.redis.sub"
    127.0.0.1:6379>
    
  2. Run the Redis Publisher

    run_redis_publisher.py
    2017-12-09 11:46:39,743 - run-redis-publisher - INFO - Start - run-redis-publisher
    2017-12-09 11:46:39,743 - run-redis-publisher - INFO - Sending msg={'account_id': 123, 'created': '2017-12-09T11:46:39.743636'} ex=reporting.accounts rk=reporting.accounts
    2017-12-09 11:46:39,767 - redis-publisher - INFO - SEND - exch=reporting.accounts rk=reporting.accounts
    2017-12-09 11:46:39,770 - run-redis-publisher - INFO - End - run-redis-publisher sent=True
    
  3. Run the Redis Kombu Subscriber

    By default, this will wait for a single message to be delivered within 10 seconds.

    kombu_redis_subscriber.py
    2017-12-09 11:47:58,798 - kombu-redis-subscriber - INFO - Start - kombu-redis-subscriber
    2017-12-09 11:47:58,798 - kombu-redis-subscriber - INFO - setup routing
    2017-12-09 11:47:58,822 - kombu-redis-subscriber - INFO - kombu-redis-subscriber - kombu.subscriber queues=reporting.accounts consuming with callback=handle_message
    2017-12-09 11:47:58,823 - kombu-redis-subscriber - INFO - callback received msg body={u'account_id': 123, u'created': u'2017-12-09T11:46:39.743636'}
    2017-12-09 11:47:58,824 - kombu-redis-subscriber - INFO - End - kombu-redis-subscriber
    
  4. Check the Redis keys

    Nothing should have changed:

    127.0.0.1:6379> keys *
    1) "_kombu.binding.reporting.accounts"
    2) "_kombu.binding.celery.redis.sub"
    127.0.0.1:6379>
    

RabbitMQ Kombu Subscriber

If you do not want to use Celery, you can use the KombuSubscriber class to process messages. This class will wait for a configurable amount of seconds to consume a single message from the subscribed queue and then stop processing.

  1. List the Queues

    If the docker containers are still running the previous RabbitMQ pub/sub test will still have the queues, exchanges and bindings still left over. If not then skip this step.

    list-queues.sh
    

    Listing Queues broker=localhost:15672

    name consumers messages messages_ready messages_unacknowledged
    celery.rabbit.sub 0 0 0 0
    reporting.accounts 0 0 0 0
  2. Run the RabbitMQ Subscriber

    Please note this output assumes there are no messages in the queue already from a previous test. It will wait for 10 seconds before stopping.

    kombu_rabbitmq_subscriber.py
    2017-12-09 11:53:56,948 - kombu-rabbitmq-subscriber - INFO - Start - kombu-rabbitmq-subscriber
    2017-12-09 11:53:56,948 - kombu-rabbitmq-subscriber - INFO - setup routing
    2017-12-09 11:53:56,973 - kombu-rabbitmq-subscriber - INFO - kombu-rabbitmq-subscriber - kombu.subscriber queues=reporting.accounts consuming with callback=handle_message
    2017-12-09 11:54:06,975 - kombu-rabbitmq-subscriber - INFO - End - kombu-rabbitmq-subscriber
    

    Or with docker compose:

    docker-compose -f compose-kombu-rabbitmq-subscriber.yml up
    Recreating kombusubrmq ... done
    Attaching to kombusubrmq
    kombusubrmq    | 2017-12-15 07:51:35,444 - kombu-rabbitmq-subscriber - INFO - Start - kombu-rabbitmq-subscriber
    kombusubrmq    | 2017-12-15 07:51:35,445 - kombu-subscriber - INFO - setup routing
    kombusubrmq    | 2017-12-15 07:51:35,479 - kombu-subscriber - INFO - kombu-rabbitmq-subscriber - kombu.subscriber queues=reporting.accounts consuming with callback=handle_message
    kombusubrmq    | 2017-12-15 07:51:45,489 - kombu-rabbitmq-subscriber - INFO - End - kombu-rabbitmq-subscriber
    kombusubrmq exited with code 0
    
  3. Run the RabbitMQ Publisher

    run_rabbitmq_publisher.py
    2017-12-09 11:56:42,793 - run-rabbitmq-publisher - INFO - Start - run-rabbitmq-publisher
    2017-12-09 11:56:42,793 - run-rabbitmq-publisher - INFO - Sending msg={'account_id': 456, 'created': '2017-12-09T11:56:42.793819'} ex=reporting rk=reporting.accounts
    2017-12-09 11:56:42,812 - rabbitmq-publisher - INFO - SEND - exch=reporting rk=reporting.accounts
    2017-12-09 11:56:42,814 - run-rabbitmq-publisher - INFO - End - run-rabbitmq-publisher sent=True
    

    Or with docker compose:

    docker-compose -f compose-run-rabbitmq-publisher.yml up
    Starting kombupubrmq ... done
    Attaching to kombupubrmq
    kombupubrmq    | 2017-12-15 07:51:50,931 - run-rabbitmq-publisher - INFO - Start - run-rabbitmq-publisher
    kombupubrmq    | 2017-12-15 07:51:50,932 - run-rabbitmq-publisher - INFO - Sending msg={'account_id': 456, 'created': '2017-12-15T07:51:50.932501'} ex=reporting rk=reporting.accounts
    kombupubrmq    | 2017-12-15 07:51:50,958 - kombu-publisher - INFO - SEND - exch=reporting rk=reporting.accounts
    kombupubrmq    | 2017-12-15 07:51:50,960 - run-rabbitmq-publisher - INFO - End - run-rabbitmq-publisher sent=True
    kombupubrmq exited with code 0
    
  4. Run the RabbitMQ Kombu Subscriber

    By default, this will wait for a single message to be delivered within 10 seconds.

    kombu_rabbitmq_subscriber.py
    2017-12-09 11:57:07,047 - kombu-rabbitmq-subscriber - INFO - Start - kombu-rabbitmq-subscriber
    2017-12-09 11:57:07,047 - kombu-rabbitmq-subscriber - INFO - setup routing
    2017-12-09 11:57:07,103 - kombu-rabbitmq-subscriber - INFO - kombu-rabbitmq-subscriber - kombu.subscriber queues=reporting.accounts consuming with callback=handle_message
    2017-12-09 11:57:07,104 - kombu-rabbitmq-subscriber - INFO - callback received msg body={u'account_id': 456, u'created': u'2017-12-09T11:56:42.793819'}
    2017-12-09 11:57:07,104 - kombu-rabbitmq-subscriber - INFO - End - kombu-rabbitmq-subscriber
    

    Or with docker compose:

    docker-compose -f compose-kombu-rabbitmq-subscriber.yml up
    Starting kombusubrmq ... done
    Attaching to kombusubrmq
    kombusubrmq    | 2017-12-15 07:51:55,366 - kombu-rabbitmq-subscriber - INFO - Start - kombu-rabbitmq-subscriber
    kombusubrmq    | 2017-12-15 07:51:55,367 - kombu-subscriber - INFO - setup routing
    kombusubrmq    | 2017-12-15 07:51:55,422 - kombu-subscriber - INFO - kombu-rabbitmq-subscriber - kombu.subscriber queues=reporting.accounts consuming with callback=handle_message
    kombusubrmq    | 2017-12-15 07:51:55,423 - kombu-rabbitmq-subscriber - INFO - callback received msg body={'account_id': 456, 'created': '2017-12-15T07:51:50.932501'}
    kombusubrmq    | 2017-12-15 07:51:55,424 - kombu-rabbitmq-subscriber - INFO - End - kombu-rabbitmq-subscriber
    kombusubrmq exited with code 0
    

Running a Redis Message Processor

This will simulate setting up a processor that handles user conversion events using a Redis server.

  1. Start the User Conversion Event Processor

    start-kombu-message-processor-redis.py
    2017-12-09 12:09:14,329 - loader-name - INFO - Start - msg-proc
    2017-12-09 12:09:14,329 - msg-proc - INFO - msg-proc START - consume_queue=user.events.conversions rk=None
    2017-12-09 12:09:14,329 - msg-sub - INFO - setup routing
    2017-12-09 12:09:14,351 - msg-sub - INFO - msg-sub - kombu.subscriber queues=user.events.conversions consuming with callback=process_message
    

    Or with docker compose:

    docker-compose -f compose-kombu-message-processor-redis.yml up
    Creating kombumsgprocredis ... done
    Attaching to kombumsgprocredis
    kombumsgprocredis    | 2017-12-15 07:54:24,167 - loader-name - INFO - Start - msg-proc
    kombumsgprocredis    | 2017-12-15 07:54:24,168 - message-processor - INFO - msg-proc START - consume_queue=user.events.conversions rk=None callback=process_message
    
  2. Publish a User Conversion Event

    From another terminal, publish a user conversion event

    publish-user-conversion-events-redis.py
    2017-12-09 12:09:16,557 - publish-user-conversion-events - INFO - Start - publish-user-conversion-events
    2017-12-09 12:09:16,558 - publish-user-conversion-events - INFO - Sending user conversion event msg={'subscription_id': 456, 'created': '2017-12-09T12:09:16.558462', 'stripe_id': 789, 'account_id': 123, 'product_id': 'ABC'} ex=user.events rk=user.events.conversions
    2017-12-09 12:09:16,582 - publish-uce-redis - INFO - SEND - exch=user.events rk=user.events.conversions
    2017-12-09 12:09:16,585 - publish-user-conversion-events - INFO - End - publish-user-conversion-events sent=True
    

    Or with docker compose:

    docker-compose -f compose-publish-user-conversion-events-redis.yml up
    WARNING: Found orphan containers (kombumsgprocredis) for this project. If you removed or renamed this service in your compose file, you can run this command with the --remove-orphans flag to clean it up.
    Creating ucepubredis ... done
    Attaching to ucepubredis
    ucepubredis    | 2017-12-15 07:54:40,539 - publish-user-conversion-events - INFO - Start - publish-user-conversion-events
    ucepubredis    | 2017-12-15 07:54:40,539 - publish-user-conversion-events - INFO - Sending user conversion event msg={'account_id': 123, 'subscription_id': 456, 'stripe_id': 789, 'product_id': 'ABC', 'created': '2017-12-15T07:54:40.539324'} ex=user.events rk=user.events.conversions
    ucepubredis    | 2017-12-15 07:54:40,619 - kombu-publisher - INFO - SEND - exch=user.events rk=user.events.conversions
    ucepubredis    | 2017-12-15 07:54:40,623 - publish-user-conversion-events - INFO - End - publish-user-conversion-events sent=True
    ucepubredis exited with code 0
    
  3. Confirm the Processor handled the conversion event

    2017-12-09 12:09:16,587 - msg-proc - INFO - msg-proc proc start - msg body={u'subscription_id': 456, u'product_id': u'ABC', u'stripe_id': 789, u'account_id': 123, u'created': u'2017-12-09T12:09:16.558462'}
    2017-12-09 12:09:16,587 - msg-proc - INFO - No auto-caching or pub-hook set exchange=None
    2017-12-09 12:09:16,588 - msg-proc - INFO - msg-proc proc done - msg
    

    Or with the docker compose version should log:

    kombumsgprocredis    | 2017-12-15 07:54:24,167 - loader-name - INFO - Start - msg-proc
    kombumsgprocredis    | 2017-12-15 07:54:24,168 - message-processor - INFO - msg-proc START - consume_queue=user.events.conversions rk=None callback=process_message
    kombumsgprocredis    | 2017-12-15 07:54:24,168 - kombu-subscriber - INFO - setup routing
    kombumsgprocredis    | 2017-12-15 07:54:40,625 - message-processor - INFO - msg-proc proc start - msg body={'account_id': 123, 'subscription_id': 456, 'stripe_id': 789, 'product_id': 'ABC', 'created': '2017-12-15T07:54:40.539324'}
    kombumsgprocredis    | 2017-12-15 07:54:40,627 - message-processor - INFO - No auto-caching or pub-hook set exchange=None
    
  4. Check the Redis keys for the new User Conversion Events key

    redis-cli
    127.0.0.1:6379> keys *
    1) "_kombu.binding.reporting.accounts"
    2) "_kombu.binding.user.events"
    3) "_kombu.binding.celery.redis.sub"
    4) "_kombu.binding.user.events.conversions"
    127.0.0.1:6379>
    

Run a Message Processor from RabbitMQ with Relay Publish Hook to Redis

This could also be set up for auto-caching instead of this pub-sub flow because this delivers a post-processing json dictionary into a Redis key (publish hook), and let’s be honest Redis is great at caching all the datas.

  1. Clear out the reporting.accounts Redis key

    Either run kombu_redis_subscriber.py until there’s no more messages being consumed or you can restart the docker containers with the stop-redis-and-rabbitmq.sh and start-redis-and-rabbitmq.sh, but the point is verify there’s nothing in the reporting.accounts key (could just delete it with the redis-cli).

  2. Start the Kombu RabbitMQ Message Processor

    start-kombu-message-processor-rabbitmq.py
    2017-12-09 12:25:09,962 - loader-name - INFO - Start - msg-proc
    2017-12-09 12:25:09,962 - msg-proc - INFO - msg-proc START - consume_queue=user.events.conversions rk=reporting.accounts
    2017-12-09 12:25:09,962 - msg-sub - INFO - setup routing
    2017-12-09 12:25:09,987 - msg-sub - INFO - msg-sub - kombu.subscriber queues=user.events.conversions consuming with callback=process_message
    

    Docker compose can start this too:

    docker stop worker;docker rm worker;
    docker-compose -f compose-kombu-message-processor-rabbitmq.yml up
    
  3. Send a User Conversion Event to RabbitMQ

    publish-user-conversion-events-rabbitmq.py
    2017-12-09 12:25:35,167 - publish-user-conversion-events - INFO - Start - publish-user-conversion-events
    2017-12-09 12:25:35,167 - publish-user-conversion-events - INFO - Sending user conversion event msg={'subscription_id': 888, 'created': '2017-12-09T12:25:35.167891', 'stripe_id': 999, 'account_id': 777, 'product_id': 'XYZ'} ex=user.events rk=user.events.conversions
    2017-12-09 12:25:35,185 - publish-uce-rabbitmq - INFO - SEND - exch=user.events rk=user.events.conversions
    2017-12-09 12:25:35,187 - publish-user-conversion-events - INFO - End - publish-user-conversion-events sent=True
    
  4. Verify the Kombu RabbitMQ Message Processor Handled the Message

    Notice the pub-hook shows the relay-specific log lines

    2017-12-09 12:25:35,188 - msg-proc - INFO - msg-proc proc start - msg body={u'subscription_id': 888, u'product_id': u'XYZ', u'stripe_id': 999, u'account_id': 777, u'created': u'2017-12-09T12:25:35.167891'}
    2017-12-09 12:25:35,188 - msg-proc - INFO - msg-proc pub-hook - build - hook msg body
    2017-12-09 12:25:35,188 - msg-proc - INFO - msg-proc pub-hook - send - exchange=reporting.accounts rk=reporting.accounts sz=json
    2017-12-09 12:25:35,210 - msg-pub - INFO - SEND - exch=reporting.accounts rk=reporting.accounts
    2017-12-09 12:25:35,212 - msg-proc - INFO - msg-proc pub-hook - send - done exchange=reporting.accounts rk=reporting.accounts res=True
    2017-12-09 12:25:35,212 - msg-proc - INFO - msg-proc proc done - msg
    
  5. Process the Redis reporting.accounts queue

    This could also be cached data about the user that made this purchase like a write-through-cache.

    kombu_redis_subscriber.py
    2017-12-09 12:26:21,846 - kombu-redis-subscriber - INFO - Start - kombu-redis-subscriber
    2017-12-09 12:26:21,846 - kombu-redis-subscriber - INFO - setup routing
    2017-12-09 12:26:21,867 - kombu-redis-subscriber - INFO - kombu-redis-subscriber - kombu.subscriber queues=reporting.accounts consuming with callback=handle_message
    2017-12-09 12:26:21,869 - kombu-redis-subscriber - INFO - callback received msg body={u'data': {}, u'org_msg': {u'subscription_id': 888, u'created': u'2017-12-09T12:25:35.167891', u'stripe_id': 999, u'product_id': u'XYZ', u'account_id': 777}, u'hook_created': u'2017-12-09T12:25:35.188420', u'version': 1, u'source': u'msg-proc'}
    2017-12-09 12:26:21,870 - kombu-redis-subscriber - INFO - End - kombu-redis-subscriber
    

SQS - Experimental

I have opened a PR for fixing the kombu http client.

  1. Export your AWS Key and Secret Key

    export SQS_AWS_ACCESS_KEY=<ACCESS KEY>
    export SQS_AWS_SECRET_KEY=<SECRET KEY>
    
  2. Publish to SQS

    kombu_sqs_publisher.py
    2017-12-09 12:49:24,900 - kombu-sqs-publisher - INFO - Start - kombu-sqs-publisher
    2017-12-09 12:49:24,901 - kombu-sqs-publisher - INFO - Sending user conversion event msg={'subscription_id': 222, 'product_id': 'DEF', 'stripe_id': 333, 'account_id': 111, 'created': '2017-12-09T12:49:24.901513'} ex=test1 rk=test1
    2017-12-09 12:49:25,007 - botocore.vendored.requests.packages.urllib3.connectionpool - INFO - Starting new HTTPS connection (1): queue.amazonaws.com
    2017-12-09 12:49:25,538 - botocore.vendored.requests.packages.urllib3.connectionpool - INFO - Starting new HTTPS connection (1): queue.amazonaws.com
    2017-12-09 12:49:26,237 - kombu-sqs-publisher - INFO - SEND - exch=test1 rk=test1
    2017-12-09 12:49:26,352 - kombu-sqs-publisher - INFO - End - kombu-sqs-publisher sent=True
    

    Or with docker compose:

    docker-compose -f compose-kombu-sqs-publisher.yml up
    
  3. Subscribe to SQS

    Please see the debugging section for getting this to function with kombu 4.1.0

    https://github.com/jay-johnson/celery-connectors#temporary-fix-for-kombu-sqs

    kombu_sqs_subscriber.py
    2017-12-09 12:49:41,232 - kombu-sqs-subscriber - INFO - Start - kombu-sqs-subscriber
    2017-12-09 12:49:41,232 - kombu-sqs-subscriber - INFO - setup routing
    2017-12-09 12:49:41,333 - botocore.vendored.requests.packages.urllib3.connectionpool - INFO - Starting new HTTPS connection (1): queue.amazonaws.com
    2017-12-09 12:49:41,801 - botocore.vendored.requests.packages.urllib3.connectionpool - INFO - Starting new HTTPS connection (1): queue.amazonaws.com
    2017-12-09 12:49:42,517 - kombu-sqs-subscriber - INFO - kombu-sqs-subscriber - kombu.subscriber queues=test1 consuming with callback=handle_message
    2017-12-09 12:49:42,671 - kombu-sqs-subscriber - INFO - callback received msg body={u'subscription_id': 222, u'created': u'2017-12-09T12:49:24.901513', u'stripe_id': 333, u'product_id': u'DEF', u'account_id': 111}
    2017-12-09 12:49:42,773 - kombu-sqs-subscriber - INFO - End - kombu-sqs-subscriber
    

    Or with docker compose:

    docker-compose -f compose-kombu-sqs-subscriber.yml up
    
  4. Verify the SQS Queue test1 is empty

    aws sqs receive-message --queue-url https://queue.amazonaws.com/<YOUR QUEUE ID>/test1
    echo $?
    0
    

Simple Pub Sub with an Existing Celery Task

Start the Celery Worker as an Ecommerce Subscriber

Please run this from the base directory of the repository in a terminal that has sourced the virtual env: source venv/bin/activate.

./start-ecomm-worker.sh

-------------- celery@ecommerce_subscriber v4.1.0 (latentcall)
---- **** -----
--- * ***  * -- Linux-4.7.4-200.fc24.x86_64-x86_64-with-fedora-24-Twenty_Four 2017-12-10 15:12:11
-- * - **** ---
- ** ---------- [config]
- ** ---------- .> app:         ecommerce-worker:0x7fae3cfa2198
- ** ---------- .> transport:   amqp://rabbitmq:**@localhost:5672//
- ** ---------- .> results:     rpc://
- *** --- * --- .> concurrency: 4 (prefork)
-- ******* ---- .> task events: OFF (enable -E to monitor tasks in this worker)
--- ***** -----
-------------- [queues]
                .> celery           exchange=celery(direct) key=celery


[tasks]
. ecomm_app.ecommerce.tasks.handle_user_conversion_events

[2017-12-10 15:12:11,727: INFO/MainProcess] Connected to amqp://rabbitmq:**@127.0.0.1:5672//
[2017-12-10 15:12:11,740: INFO/MainProcess] mingle: searching for neighbors
[2017-12-10 15:12:12,776: INFO/MainProcess] mingle: all alone
[2017-12-10 15:12:12,828: INFO/MainProcess] celery@ecommerce_subscriber ready.
[2017-12-10 15:12:13,633: INFO/MainProcess] Events of group {task} enabled by remote.

Publish User Conversion Events to the Celery Ecommerce Subscriber

Please run this from a separate terminal that has sourced the virtual env: source venv/bin/activate.

  1. Change to the ecomm_app directory

    cd ecomm_app
    
  2. Publish a task

    This will use the Celery send_task method to publish the Celery task: ecomm_app.ecommerce.tasks.handle_user_conversion_events to RabbitMQ which is monitored by the Celery ecommerce worker.

    ./publish_task.py
    INFO:celery-task-publisher:Sending broker=amqp://rabbitmq:rabbitmq@localhost:5672// body={'subscription_id': 321, 'msg_id': '6d7ab602-f7cd-4d90-a0c5-5eb0cdcb41d9', 'version': 1, 'product_id': 'JJJ', 'account_id': 999, 'stripe_id': 876, 'created': '2017-12-10T15:16:08.557804'}
    INFO:celery-task-publisher:Done with msg_id=6d7ab602-f7cd-4d90-a0c5-5eb0cdcb41d9 result=True
    

Confirm the Celery Worker Processed the Conversion Message

[2017-12-10 15:16:08,593: INFO/MainProcess] Received task: ecomm_app.ecommerce.tasks.handle_user_conversion_events[9349e1be-fca5-40b5-86d3-0661fdd9fd06]
[2017-12-10 15:16:08,594: INFO/ForkPoolWorker-4] task - user_conversion_events - start body={'stripe_id': 876, 'version': 1, 'subscription_id': 321, 'created': '2017-12-10T15:16:08.557804', 'account_id': 999, 'product_id': 'JJJ', 'msg_id': '6d7ab602-f7cd-4d90-a0c5-5eb0cdcb41d9'}
[2017-12-10 15:16:08,595: INFO/ForkPoolWorker-4] task - user_conversion_events - done
[2017-12-10 15:16:08,619: INFO/ForkPoolWorker-4] Task ecomm_app.ecommerce.tasks.handle_user_conversion_events[9349e1be-fca5-40b5-86d3-0661fdd9fd06] succeeded in 0.025004414000250108s: True

Check the Ecommerce Subscriber in Flower

The Ecommerce Publisher and Subscriber are using RabbitMQ which is registered under the Flower url:

http://localhost:5555/ - (login: admin/admin)

There should be a Worker named:

celery@ecommerce_subscriber

There are also additional worker details available at:

http://localhost:5555/worker/celery@ecommerce_subscriber

View the registered ecommerce tasks for the worker:

http://localhost:5555/worker/celery@ecommerce_subscriber#tab-tasks

Debugging with rabbitmqadmin

The pip and development build will install rabbitmqadmin (https://raw.githubusercontent.com/rabbitmq/rabbitmq-management/v3.7.0/bin/rabbitmqadmin) version 3.7.0. It is a great utility for verifying RabbitMQ messaging and does not require having access to the RabbitMQ cluster’s host nodes (or a machine with rabbitmqctl on it).

Please note: rabbitmqadmin uses the management HTTP port (not the amqp port 5672) which requires a broker to have the management plugin enabled to work if you’re using this with an external RabbitMQ cluster.

Checking queues

Script in pip

list-queues.sh

Listing Queues broker=localhost:15672

+--------------------+-----------+----------+----------------+-------------------------+
|        name        | consumers | messages | messages_ready | messages_unacknowledged |
+--------------------+-----------+----------+----------------+-------------------------+
| celery             | 0         | 0        | 0              | 0                       |
| reporting.accounts | 0         | 0        | 0              | 0                       |
+--------------------+-----------+----------+----------------+-------------------------+

Manual way

rabbitmqadmin.py --host=localhost --port=15672 --username=rabbitmq --password=rabbitmq list queues
+--------------------+-----------+----------+----------------+-------------------------+
|        name        | consumers | messages | messages_ready | messages_unacknowledged |
+--------------------+-----------+----------+----------------+-------------------------+
| celery             | 0         | 0        | 0              | 0                       |
| reporting.accounts | 0         | 0        | 0              | 0                       |
+--------------------+-----------+----------+----------------+-------------------------+

Checking exchanges

Script in pip

list-exchanges.sh

Listing Exchanges broker=localhost:15672

+---------------------+---------+
|        name         |  type   |
+---------------------+---------+
|                     | direct  |
| amq.direct          | direct  |
| amq.fanout          | fanout  |
| amq.headers         | headers |
| amq.match           | headers |
| amq.rabbitmq.log    | topic   |
| amq.rabbitmq.trace  | topic   |
| amq.topic           | topic   |
| celery              | direct  |
| celery.pidbox       | fanout  |
| celeryev            | topic   |
| reply.celery.pidbox | direct  |
| reporting.accounts  | topic   |
+---------------------+---------+

Manual way

rabbitmqadmin.py --host=localhost --port=15672 --username=rabbitmq --password=rabbitmq list exchanges name typa
+---------------------+---------+
|        name         |  type   |
+---------------------+---------+
|                     | direct  |
| amq.direct          | direct  |
| amq.fanout          | fanout  |
| amq.headers         | headers |
| amq.match           | headers |
| amq.rabbitmq.log    | topic   |
| amq.rabbitmq.trace  | topic   |
| amq.topic           | topic   |
| celery              | direct  |
| celery.pidbox       | fanout  |
| celeryev            | topic   |
| reply.celery.pidbox | direct  |
| reporting.accounts  | topic   |
+---------------------+---------+

List Bindings

Script in pip

list-bindings.sh

Listing Bindings broker=localhost:15672

+--------------------+--------------------+--------------------+
|       source       |    destination     |    routing_key     |
+--------------------+--------------------+--------------------+
|                    | celery             | celery             |
|                    | reporting.accounts | reporting.accounts |
| celery             | celery             | celery             |
| reporting          | reporting.accounts | reporting.accounts |
+--------------------+--------------------+--------------------+

Manual way

rabbitmqadmin.py --host=localhost --port=15672 --username=rabbitmq --password=rabbitmq list bindings source destination routing_key
+--------------------+--------------------+--------------------+
|       source       |    destination     |    routing_key     |
+--------------------+--------------------+--------------------+
|                    | celery             | celery             |
|                    | reporting.accounts | reporting.accounts |
| celery             | celery             | celery             |
| reporting          | reporting.accounts | reporting.accounts |
+--------------------+--------------------+--------------------+

Development Guide

  1. Install the development environment

    virtualenv -p python3 venv && source venv/bin/activate && pip install -e .
    
  2. Run tests

    The tests require the docker containers to be running prior to starting.

    python setup.py test
    

Debugging

pycURL Reinstall with NSS

For anyone wanting to use kombu SQS, I had to uninstall pycurl and install it with nss.

The error looked like this in the logs:

2017-12-09 12:28:46,811 - kombu-sqs-subscriber - INFO - kombu-sqs-subscriber - kombu.subscriber consume hit exception=The curl client requires the pycurl library. queue=test1

So I opened up a python shell

Python 2:

$ python
Python 2.7.12 (default, Sep 29 2016, 13:30:34)
[GCC 6.2.1 20160916 (Red Hat 6.2.1-2)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import pycurl
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "build/bdist.linux-x86_64/egg/pycurl.py", line 7, in <module>
File "build/bdist.linux-x86_64/egg/pycurl.py", line 6, in __bootstrap__
ImportError: pycurl: libcurl link-time ssl backend (nss) is different from compile-time ssl backend (none/other)
>>>

Python 3:

$ python
Python 3.5.3 (default, May 11 2017, 09:10:41)
[GCC 6.3.1 20161221 (Red Hat 6.3.1-1)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import pycurl
Traceback (most recent call last):
    File "<stdin>", line 1, in <module>
ImportError: pycurl: libcurl link-time ssl backend (nss) is different from compile-time ssl backend (none/other)
>>>

Uninstalled and Reinstalled pycurl with nss

pip uninstall -y pycurl; pip install pycurl --compile --global-option="--with-nss" pycurl

Temporary fix for Kombu SQS

SQS Kombu Subscriber 'NoneType' object has no attribute 'call_repeatedly'

Until Kombu fixes the SQS transport and publishes it to pypi, the SQS subscriber will throw exceptions like below.

kombu_sqs_subscriber.py
2017-12-09 12:30:45,493 - kombu-sqs-subscriber - INFO - Start - kombu-sqs-subscriber
2017-12-09 12:30:45,493 - kombu-sqs-subscriber - INFO - setup routing
2017-12-09 12:30:45,602 - botocore.vendored.requests.packages.urllib3.connectionpool - INFO - Starting new HTTPS connection (1): queue.amazonaws.com
2017-12-09 12:30:46,046 - botocore.vendored.requests.packages.urllib3.connectionpool - INFO - Starting new HTTPS connection (1): queue.amazonaws.com
2017-12-09 12:30:46,832 - kombu-sqs-subscriber - INFO - kombu-sqs-subscriber - kombu.subscriber queues=test1 consuming with callback=handle_message
2017-12-09 12:30:46,989 - kombu-sqs-subscriber - INFO - callback received msg body={u'subscription_id': 222, u'created': u'2017-12-09T12:28:28.093582', u'stripe_id': 333, u'product_id': u'DEF', u'account_id': 111}
2017-12-09 12:30:46,994 - kombu-sqs-subscriber - INFO - kombu-sqs-subscriber - kombu.subscriber consume hit exception='NoneType' object has no attribute 'call_repeatedly' queue=test1
2017-12-09 12:30:46,994 - kombu-sqs-subscriber - INFO - End - kombu-sqs-subscriber
Restoring 1 unacknowledged message(s)

Notice the last line has put the message into SQS in-flight which means it has not been acknowledged or deleted.

You can verify this message is still there with the aws cli:

aws sqs receive-message --queue-url https://queue.amazonaws.com/<YOUR QUEUE ID>/test1
{
    "Messages": [
        {
            "Body": "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",
            "ReceiptHandle": "AQEBDnxqT1+SOam1ZtMKPgh77a8bapLbcrI3PZRTqVZJokz0h7oMusuJPAB9jksH3BQHQyg3TyZXasBblpMcin3HTzh7ykTgAgawhMreOoWGGiaeEoOekaChn2yFpKDbVP1ZENRVcpAzeDXzCd52TITZbyLk8FY1PJB3XpAiih9SH/R0FPj3JnU0WTxjTAWtBnSlUUGXFc3CczJi61YsJS+bTZs8JIgDaICMF+zMhnV+rV4zXDObTVFM3OaMdf/puqZ9yRd3fM1GsOxZaDNRDGYKml/UK0tn32gtqPSuUW905YamwnWQYB9mF338Jgx11rv78b5lLogpU/0t6E+0tD1Lkr/UR/M64NZI2eTwp6ZHNtqTNbkjd5VsBgB39b+wXFFn",
            "MD5OfBody": "e72609877b90ad86df2f161c6303eaf0",
            "MessageId": "684328b4-a38c-4868-8550-e0d46599a0c2"
        }
    ]
}

If you’re feeling bold, you can run off my PR fix branch as well:

pip uninstall -y kombu ; rm -rf /tmp/sqs-pr-fix-with-kombu; git clone https://github.com/jay-johnson/kombu.git /tmp/sqs-pr-fix-with-kombu && pushd /tmp/sqs-pr-fix-with-kombu && git checkout sqs-http-get-client && python setup.py develop && popd

With the SQS fix applied locally (works on python 2 and 3 on my fedora 24 vm):

2017-12-09 12:47:12,177 - kombu-sqs-subscriber - INFO - Start - kombu-sqs-subscriber
2017-12-09 12:47:12,177 - kombu-sqs-subscriber - INFO - setup routing
2017-12-09 12:47:12,295 - botocore.vendored.requests.packages.urllib3.connectionpool - INFO - Starting new HTTPS connection (1): queue.amazonaws.com
2017-12-09 12:47:12,736 - botocore.vendored.requests.packages.urllib3.connectionpool - INFO - Starting new HTTPS connection (1): queue.amazonaws.com
2017-12-09 12:47:13,454 - kombu-sqs-subscriber - INFO - kombu-sqs-subscriber - kombu.subscriber queues=test1 consuming with callback=handle_message
2017-12-09 12:47:13,592 - kombu-sqs-subscriber - INFO - callback received msg body={u'subscription_id': 222, u'created': u'2017-12-09T12:28:28.093582', u'stripe_id': 333, u'product_id': u'DEF', u'account_id': 111}
2017-12-09 12:47:13,689 - kombu-sqs-subscriber - INFO - End - kombu-sqs-subscriber

After running it you can confirm the message has been deleted and acknowledged with the aws cli:

aws sqs receive-message --queue-url https://queue.amazonaws.com/<YOUR QUEUE ID>/test1
echo $?
0

Testing

Start the Relay

./start-ecomm-relay.py

Start the Celery Worker

./start-ecomm-worker.sh

Load Test Celery Worker over RabbitMQ

This will send 50,000 messages over with the Celery send_task method. As long as the ecomm Celery worker is running the messages will be sent over.

python -m unittest tests/load_test_worker_rabbitmq.py

Load Test Relay

This will send 50,000 messages over the user.events.conversions RabbitMQ queue for the ecomm relay to process and then send to the ecomm worker.

python -m unittest tests/load_test_relay_rabbitmq.py

Cleanup Persistence

Docker compose creates files and directories as the host’s root user. This makes cleaning up test runs annoying. So here’s a tool to clean the persistence data and logs but it requires providing sudo or running as root.

Please, please, please be careful!

  1. Shut them down to prevent writes to the volumes

    stop-redis-and-rabbitmq.sh
    
  2. Clean them up

    sudo ./clean-persistence-data.sh
    
    Using root to delete persistence directories: ./docker/data/rabbitmq/ ./docker/data/redis and logs: ./docker/logs/rabbitmq ./docker/logs/redis and files: ./docker/data/rabbitmq/.erlang.cookie
    
    - deleting=./docker/data/rabbitmq
    - deleting=./docker/logs/rabbitmq
    - deleting=./docker/data/redis
    - deleting=./docker/logs/redis
    - deleting=./docker/data/rabbitmq/.erlang.cookie
    
  3. Start them up again

    start-persistence-containers.sh
    

Create your own self-signed Keys, Certs and Certificate Authority with Ansible

If you have openssl installed you can use this ansible playbook to create your own certificate authority (CA), keys and certs.

  1. Create the CA, Keys and Certificates

    cd ansible
    ansible-playbook -i inventory_dev create-x509s.yml
    
  2. Verify the Jupyter Client Cert

    openssl x509 -in ../compose/ssl/client_cert.pem -text -noout
    
  3. Verify the Jupyter Server Cert

    openssl x509 -in ../compose/ssl/jupyter_server_cert.pem -text -noout
    
  4. Using the certs

    Docker makes testing ssl easier so the certs are created under the compose/ssl directory:

    tree ../compose/ssl
    ├── ca.pem
    ├── ca_private_key.pem
    ├── client_cert.pem
    ├── client.csr
    ├── client_key.pem
    ├── database_server_cert.pem
    ├── database_server.csr
    ├── database_server_key.pem
    ├── docker_server_cert.pem
    ├── docker_server.csr
    ├── docker_server_key.pem
    ├── extfile.cnf
    ├── jenkins_server_cert.pem
    ├── jenkins_server.csr
    ├── jenkins_server_key.pem
    ├── jupyter_server_cert.pem
    ├── jupyter_server.csr
    ├── jupyter_server_key.pem
    ├── kibana_server_cert.pem
    ├── kibana_server.csr
    ├── kibana_server_key.pem
    ├── nginx_server_cert.pem
    ├── nginx_server.csr
    ├── nginx_server_key.pem
    ├── pgadmin_server_cert.pem
    ├── pgadmin_server.csr
    ├── pgadmin_server_key.pem
    ├── phpmyadmin_server_cert.pem
    ├── phpmyadmin_server.csr
    ├── phpmyadmin_server_key.pem
    ├── rabbitmq_server_cert.pem
    ├── rabbitmq_server.csr
    ├── rabbitmq_server_key.pem
    ├── redis_server_cert.pem
    ├── redis_server.csr
    ├── redis_server_key.pem
    ├── restapi_server_cert.pem
    ├── restapi_server.csr
    ├── restapi_server_key.pem
    ├── webserver_server_cert.pem
    ├── webserver_server.csr
    └── webserver_server_key.pem
    
  5. Set up your own extfile.cnf - Optional

    You can change the source extfile.cnf which is copied over to the compose/ssl directory when the playbook runs as needed.

    cat ./configs/extfile.cnf
    subjectAltName = DNS:*.localdev.com, DNS:rabbitmq.localdev.com, DNS:redis.localdev.com, DNS:jupyter.localdev.com, DNS:jenkins.localdev.com, DNS:www.localdev.com, DNS:api.localdev.com, DNS:db.localdev.com, DNS:pgadmin.localdev.com, DNS:phpmyadmin.localdev.com, DNS:kibana.localdev.com, DNS:lb.localdev.com, DNS:docker.localdev.com, IP:127.0.0.1
    extendedKeyUsage = serverAuth
    
  6. Customizing your own openssl.cnf and cert_openssl.cnf - Optional

    You can change the source openssl.cnf before creating the certs.

    cat ./configs/openssl.cnf
    [ req ]
    prompt              = no
    default_bits        = 2048
    distinguished_name  = req_distinguished_name # where to get DN for reqs
    
    [ req_distinguished_name ]
    C  = US
    ST = WA
    L  = Redmond
    O  = SecureEverything
    OU = SecureEverythingOrgUnit
    CN = LocalDev
    

    You can change the source cert_openssl.cnf before creating the certs.

    cat ./configs/cert_openssl.cnf
    [req]
    days                   = 2000
    serial                 = 1
    distinguished_name     = req_distinguished_name
    x509_extensions        = v3_ca
    
    
    [req_distinguished_name]
    countryName            = US
    stateOrProvinceName    = WA
    localityName           = Redmond
    organizationName       = SecureEverything
    organizationalUnitName = SecureEverythingOrgUnit
    commonName             = SecureEverything
    
    [ v3_ca ]
    subjectKeyIdentifier   = hash
    authorityKeyIdentifier = keyid:always,issuer:always
    basicConstraints       = CA:TRUE
    keyUsage               = digitalSignature, nonRepudiation, keyEncipherment, dataEncipherment, keyAgreement, keyCertSign
    subjectAltName         = DNS:*.localdev.com, DNS:redis.localdev.com, DNS:rabbitmq.localdev.com, DNS:jupyter.localdev.com, DNS:jenkins.localdev.com, DNS:www.localdev.com, DNS:api.localdev.com, DNS:db.localdev.com, DNS:pgadmin.localdev.com, DNS:phpmyadmin.localdev.com, DNS:kibana.localdev.com, DNS:lb.localdev.com, DNS:docker.localdev.com, email:admin@localdev.com
    issuerAltName          = issuer:copy
    

    I found this link helpful for understanding all the different configurable options: https://www.ibm.com/support/knowledgecenter/en/SSB23S_1.1.0.13/gtps7/cfgcert.html

Running JupyterHub with Postgres and SSL

  1. Pull the default Jupyter image

    All users will share this large 4.4 gb image

    docker pull jupyter/scipy-notebook:latest
    
  2. Append the following entries to your /etc/hosts row with 127.0.0.1

    jupyter.localdev.com rabbitmq.localdev.com redis.localdev.com jenkins.localdev.com

  3. Verify /etc/hosts has the entries

    cat /etc/hosts | grep localdev
    127.0.0.1      localhost localhost.localdomain localhost4 localhost4.localdomain4 jupyter.localdev.com rabbitmq.localdev.com redis.localdev.com jenkins.localdev.com
    
  4. From the base repository directory, change to the compose directory

    cd compose
    
  5. Create the JupyterHub docker network

    This should only be required if the jupyterhub-network does not already exist.

    docker network create jupyterhub-network
    
  6. Create the JupyterHub docker data volume

    This should only be required if the jupyterhub-data does not already exist.

    docker volume create --name jupyterhub-data
    

    Each user will need a data volume if they are not already created as well with naming scheme:

    jupyterhub-user-<username> to persist notebooks.

    docker volume create --name jupyterhub-user-admin
    
  7. Start JupyterHub

    docker stop jupyterhub ; docker rm jupyterhub; docker-compose -f compose-jupyter.yml up
    
  8. Login to JupyterHub

    Please change these defaults before deploying to production:

    • username: admin
    • password: admin

    Please accept to “Proceed” passed the self-signed certificate warning.

    https://jupyter.localdev.com/hub/login

    https://raw.githubusercontent.com/jay-johnson/celery-connectors/master/_images/jupyterhub-step-1-login-as-admin-admin.png
  9. Start the Admin user Jupyter instance

    Click on Start My Server

    https://raw.githubusercontent.com/jay-johnson/celery-connectors/master/_images/jupyterhub-step-2-start-server.png
  10. Clone some great notebooks into the Admin Jupyter workspace

    From a terminal with access to docker clone a repository with some amazing ipython notebooks:

    https://github.com/donnemartin/data-science-ipython-notebooks

    docker exec -it jupyter-admin git clone https://github.com/donnemartin/data-science-ipython-notebooks.git /home/jovyan/work/data-science-ipython-notebooks
    Cloning into '/home/jovyan/work/data-science-ipython-notebooks'...
    remote: Counting objects: 2344, done.
    remote: Total 2344 (delta 0), reused 0 (delta 0), pack-reused 2344
    Receiving objects: 100% (2344/2344), 47.76 MiB | 16.95 MiB/s, done.
    Resolving deltas: 100% (1317/1317), done.
    Checking connectivity... done.
    
  11. Browse the cloned notebooks

    https://jupyter.localdev.com/user/admin/tree/work/data-science-ipython-notebooks

    https://raw.githubusercontent.com/jay-johnson/celery-connectors/master/_images/jupyterhub-step-3-browse-ipython-notebooks.png
  12. Open the one of the cloned notebooks

    https://jupyter.localdev.com/user/admin/notebooks/work/data-science-ipython-notebooks/scikit-learn/scikit-learn-intro.ipynb

  13. Select Kernel -> Restart & Run All

    Confirm you can run all the cells in the notebook

  14. Verify the notebook ran all the cells without any errors

    Save the output and changes to the notebook with ctrl + s. At the bottom of the notebook you should see the updated chart for the sepal width and sepal-length similar to:

    https://raw.githubusercontent.com/jay-johnson/celery-connectors/master/_images/jupyterhub-step-4-run-all-notebook-cells.png
  15. Verify the notebook was changed and updated

    Browse to:

    https://jupyter.localdev.com/user/admin/tree/work/data-science-ipython-notebooks/scikit-learn

    The scikit-learn-intro.ipynb should be running and updated.

    https://raw.githubusercontent.com/jay-johnson/celery-connectors/master/_images/jupyterhub-step-5-confirm-notebook-was-saved.png
  16. Stop the Admin Jupyter instance

    The notebooks should persist a stop and start of a user’s Jupyter container instance.

    https://jupyter.localdev.com/hub/admin

    It should look something like this:

    https://raw.githubusercontent.com/jay-johnson/celery-connectors/master/_images/jupyterhub-step-6-stop-server.png
  17. Start the Admin Jupyter instance again

    Click start server

  18. Browse to the scikit-learn directory and confirm the files were not lost on the restart

    https://jupyter.localdev.com/user/admin/tree/work/data-science-ipython-notebooks/scikit-learn

    https://raw.githubusercontent.com/jay-johnson/celery-connectors/master/_images/jupyterhub-step-7-jupyterhub-user-notebook-persistence.png

Linting

pycodestyle --max-line-length=160 --exclude=venv,build,.tox,celery_connectors/rabbitmq/rabbitmqadmin.py

License

Apache 2.0 - Please refer to the LICENSE for more details

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for celery-connectors, version 1.0.30
Filename, size File type Python version Upload date Hashes
Filename, size celery_connectors-1.0.30-py2.py3-none-any.whl (94.2 kB) File type Wheel Python version py2.py3 Upload date Hashes View
Filename, size celery-connectors-1.0.30.tar.gz (110.5 kB) File type Source Python version None Upload date Hashes View

Supported by

Pingdom Pingdom Monitoring Google Google Object Storage and Download Analytics Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page