Skip to content

Latest commit

 

History

History
140 lines (119 loc) · 5.96 KB

CHANGELOG.md

File metadata and controls

140 lines (119 loc) · 5.96 KB
Raw

Version 0.2.1 - 2018-12-07

  • Fix a bug causing crashes when using remote healthchecks

Version 0.2.0 - 2018-12-06

  • Upgrade to build with Go 1.11
  • Support multiple NICs on a machine
  • Check EC2 instance healthchecks as well as route blackholing so that routes wil be replaced if held by an instance that is impaired as well as just terminated.
  • Fix race condition with healthchecks so that routes cannot be taken over before the healthcheck has passed.

Version 0.1.5 - 2017-05-19

  • Add by_tag_regexp matcher to the route tables finder (ashb)
  • Upgrade to build with Go 1.8

Version 0.1.4 - 2017-05-19

  • Drop Gom due to it being more hassle than it's worth
  • Use foked version of logarus as sirupsen changed his github account name (breaking all imports) without updating all of the logarus code, so upstream is currently unuseable
  • Upgrade to build with Go 1.7

Version 0.1.3 - 2016-12-19

  • Add version to the user agent sent.
  • Upgrade AWS SDK
  • Additional paranoia and debug logging

Version 0.1.2 - 2016-04-15

  • Initial support for running hooks on creating/deleting routes. Not yet documented or fully tested, but this is the first part of fixing issue #1
  • Switch from Godep to Gom to manage dependencies.
  • Upgrade aws-sdk-go to the latest upstream version.

Version 0.1.1 - 2016-03-05

  • Add a 'command' healthcheck type.
  • Documentation updates / improvements
  • Better type conversions in the "config" attribute for health checks (Fixes #6)

Version 0.1.0 - 2016-02-01

  • Add the ability to negotiate SSL connections when doing healthchecks.
  • Remove some of the scary warning text from README

Version 0.0.10 - 2015-12-05

  • First version with working remote_healthchecks

Version 0.0.9 - 2015-12-04

  • Fix various crash cases
  • Almost working remote_healthchecks
  • More improvements to error reporting in configuration

Version 0.0.8 - 2015-11-29

  • Fix crash on route tables which don't have an instance-id attached.
  • More improvements to error reporting in configuration
  • Make the error reporting of healthcheck configuration better by reporting all errors at once, rather than the first one.

Version 0.0.7 - 2015-11-28 BROKEN DO NOT USE

  • Make config tell you about all the errors found when dieing at startup, rather than just the first one found.
  • Now requires the ec2:DescribeInstanceAttribute permission to run.
  • Local instance is checked for src/destination check being disabled at startup. If not disabled AWSnycast will quit with an error message (as you need this disabled for either NAT boxes or those handinging Anycast addresses)
  • Add a -syslog CLI flag, which will cause the logs printed on STDOUT to be duplicated to syslog.
  • When replacing an existing route, log the old and new instance IDs, and the route state.
  • Improved unit tests - added coverage.
  • Additional work towards the remote_healthchecks feature working.

Version 0.0.6 - 2015-11-13

  • Build .deb and .rpm packages for Linux as part of the release process

Version 0.0.5 - 2015-11-13

  • Build a .deb package for amd64 as part of the release process
  • Add additional methods to find route tables
  • More logging improvements
  • Fix the integration tests to not run Terrafom themselves

Version 0.0.4 - 2015-11-13

  • Make the logging much better, with more context to log messages
    • Make debug mode output more verbose debug logging
  • Add integration tests of Anycast functionality
  • Make the integration test environment useful again
  • Start using the 'godep' tool to freeze dependencies

Version 0.0.3 - 2015-11-12

  • Add much better documentation
  • Add the ability to setup a route finder which does not bomb out it doesn't find anything

Version 0.0.2 - 2015-11-12

  • Make the interval between polls to AWS for the current routing tables be configurable with the top level poll_time key. (Defaults to 300s)
  • Add a never_delete flag which can/should be used with the default route (0.0.0.0/0) when providing NAT service to not delete the route if connectivity to a healthcheck location fails.
  • Expand route filters with two new types 'and' and 'or'. These both take a single config key, 'filters', which is a list of other filters to and/or together. If you have a few consistent tags on your route tables then you can use this to create logic for finding sets of route tables, rather than having to hard code route table names/numbers/etc into the config.
  • Update to the breaking changes in the latest version of aws-sdk-go
  • Deb packaging example added to the repository
  • Expand the integration test examples to use all the features above, and have an example of an Anycast type service and not just NAT.

Version 0.0.1 - 2015-10-25

  • Well, @garethr put this in devops weekly, so I guess I should cut a binary release.
  • Currently working feature set:
    • NAT (0.0.0.0/0) or Anycast (/32) routes can be injected into route tables found by tags, when healthchecks pass.
    • Simple ping / tcp healthchecks work. (I think maaaybe you can do HTTP with this, but haven't tried it in the integration tests yet!)
    • Routes are deleted if owned by the current instance and the healthcheck starts failing.
    • Routes are added if not present (or attached to an instance/ENI which is dead / blackholing traffic) as soon as healthchecks start passing
    • AWS is polled (DescribeRouteTables) every 300s.
  • Important things which don't work.
    • Do not try this if you have multiple VPCs in the same region, with the same tags on routing tables, you will have a bad time.
    • HA/takeover is only done for blackholed traffic (dead instance/disconnected ENI). You cannot (yet) healthcheck other instances (e.g. the current instance providing a route) - so if an instance gets a route, AWSnycast stops running on that instance, and the service breaks, then it'll not be fixed until you manually stop or terminate the unhealthy instance!