Skip to content
This repository has been archived by the owner on Aug 5, 2022. It is now read-only.

Latest commit

 

History

History
57 lines (41 loc) · 2.44 KB

release_notes.md

File metadata and controls

57 lines (41 loc) · 2.44 KB
Raw

Release Process

We release to PyPI using twine.

The process is

  1. Increment the project version in setup.py
  2. Clear the dist directory
  3. From the project root, run the setup tools to build source distribution and wheel in the dist directory
pipenv install -e .
pipenv install --dev
pipenv run python setup.py upload

Release Notes

Packages are available on PyPI

You can use the run_tests.sh file to run the test suite via docker. The script uses the tests-docker-compose.yml file to provide the dependencies for the tests, deploys the code and then runs the test suite.

Master

Release 0.6.9

-- Fixed an issue where the heartbeat was not sent if a long-running handler took to long to process a request. You can now flag a connection as having a long-running handler and we will spin off a thread to send heartbeat messages whilst the hanlder is running. -- Reviewed where OSError exceptions were being caught. Moved lower and re-raised ans a ChannelFailure exception

Release 0.6.5

-- RMQ, Fixed the issue with the parameter "x-hapolicy" which should have been "x-ha-policy". Needed to support older 2.X versions

Release 0.6.4

-- Fixed issues produced during partition testing

Assume I have a cluster with three nodes: A,B, C

Assume that I have a queue, orders, with the master on A and slaves on B and C.

We have chosen a strategy of Pause Minority on a partition

Assume I connect to Node B. I consume from the master on A via Node B. (I don't consume from the slave, that is there in case A fails). Then B gets a partition and I cannot see A or C. I need to re-connect to A via A or C. I have chosen an Pause Minority strategy. * RMQ will pause the partioned node * We will timeout on our connection * I need to stop talking to B and talk to A or C.

Results
* Consumer is Idle
    * RMQ reports that only one mirror now exists
    * Socket error on Brightside client
    * Consumer connects to new node ater delay
* Consumer is Busy
    * RMQ reports that only one mirror now exists
    * Socket error on Brightside client
    * Consumer connects to new node ater delay
        * Does not seem to consume the messages from the queue -- appears to be a zombie
        * Fixed - handle socket errors by recreating connection

-- Fixed issue that older versions of RabbitMQ do not use policy and require the x-hapolicy flag to be set at queue creation.