v5.1.1

• Fixed bootstrap server reboot #70 @Tensibai
• Fix cfn-lint errors W3011 & E3012 and update to latest AMI builds #74 @gsreynolds
• Switch push-jobs-server back to stable channel - current channel doesn't exist for push jobs server #75 @gsreynolds

5.1.0

Improvements since v5.0.4:

• Updates to Chef Server 13.1.8, Push Jobs Server 3.0.3 and Supermarket 3.3.7
• Adds the ChefServerCustomConfig parameter to main.yaml #68 @jeffj254
• Updates the ELB security policy to TLS-1-2-2017-01 #67 @jeffj254
• Add ability to enter subnets for ELBs and RDS DB #65 @jeffj254
• Switch S3 paths to use virtual-style pathing #60 @jeffj254

5.0.4

HUGE reliability improvement
This version solves a major longstanding issue faced by users in 3-node Elasticsearch configurations, where they were all forced to be placed in a single AZ and were vulnerable to AZ-level failures.

NOTE: this change will likely force a rebuild of your Elasticsearch cluster, and will force users to run the following on the bootstrap frontend:

1. chef-server-ctl reconfigure
2. chef-server-ctl reindex --all-orgs --disable-api

Fixes: #49

5.0.3

• Major reliability improvement: Make the Chef Server frontend ASG automatically rebuild instances that go unhealthy, but not bootstrap instances.
  • Add cloudwatch alerts for all unhealthy hosts.
• Add clarity about the email address to the PREREQS.md doc
• Deploy AMIs to every region
• AMIs have rolled back the Chef Server version to 12.18.x in order to avoid an LDAP SSL bug
• [ALB Target Groups] Reduce unhealthy threshold from 10 minutes to 1 minute and double the check frequency

fixes #55, fixes #54

v5 and the road to a Marketplace-like experience

V5 of this template makes a significant and breaking change from v4.

This template depends on a pre-built AMI that includes all of the needed files, thus saving users from the many problems that could occur attempting to download packages and configuration scripts. This is also a hard requirement for registration in the AWS Marketplace.

Additionally, there is a new marketplace.yaml template which aims to simplify the startup experience and also run at a significantly reduced cost. While the main template is geared for high-performance and would cost around $1500 per month, the marketplace template defaults run for about $320 per month. AWS cost-saving measures such as Reserved Instances can reduce these costs further.

To try this out for yourself, see the updated instructions in the README or try the quickstart:

Edition	Version	Est Monthly cost	Launch button
High Performance	5.0.2	$1500*
Lowest Cost	5.0.2	$320*

*NOTE: Estimated costs are just that, your mileage will vary. Use reserved instances to greatly reduce costs for long-running clusters.

Fixes

fixes #46, fixes #44, fixes #43, fixes #42, fixes #51, fixes #52, fixes #53

New Features

• Ability to supply an admin password to Automate, rather than having one generated for you
• Significantly rewritten PREREQS.md document

4.1.0

Bug-fix release:

• fix install of awscli on RHEL again
• Update OS mappings and don't exclude kernel/dracut or bad things can happen in the yum upgrade

4.0.0

The goal of the 4.0 release is to enhance the aws_native stack to also include Chef Automate 2 and Supermarket, providing the complete stack to customers.

Those new servers are provided in non-HA flavors today, however they use the EC2 System Autorecovery feature, plus self-snapshotting of data, in order to provide more than acceptable availability.

Major improvements to be excited about:

• Switch to CentOS and stock RHEL AMIs instead of Amazon Linux
• refactor cloudformation into nested stacks, launched by a single main.yaml stack
• Add a Supermarket stack (auto-recovering standalone)
• Add a Chef Automate 2 stack (auto-recovering standalone)
• Perform automatic nightly EBS snapshots of the Automate and Supermarket data volumes
• Simplify and deduplicate the IAM profiles, Security Groups and Log groups between the stacks.
• Update the before.sh script to automatically install awscli and cfn-init on stock RHEL
• Trim the allowed instance type list to T3, M5 and C5 classes because it's 2018 and we can all have nice things
• More thorough testing in travis

3.3.3

Version updates for:

• Amazon Linux: 2017.09.1.20180115 -> 2018.03.0.20180412
• Postgres RDS: 9.6.5 -> 9.6.8
• Chef Server: 12.17.15 -> 12.17.33
• Chef Manage: 2.5.8 -> 2.5.15
• Push Jobs Server: 2.2.6 -> 2.2.8

3.3.1

• Adds all of the potential Event Categories for a RDS Event Subscription
• Comments out the noisy + unneeded Event Categories: backup & read replica

basically, you won't get an alert for nightly backups anymore 😄

3.3.0

adds notifications for RDS and AutoScaling group actions, which go to the Alert notification topic (for consumption by email and/or alerting systems)