⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️

[!IMPORTANT]
Documentation Update: Product documentation previously located in /website has moved to the hashicorp/web-unified-docs repository, where all product documentation is now centralized. Please make contributions directly to web-unified-docs, since changes to /website in this repository will not appear on developer.hashicorp.com. ⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️⚠️

Upgrade to v1.12

The topic describes upgrade considerations for Packer v1.12. Refer to the v1.12 release notes for details about changes in v1.12.

Overview

You can use the upgrade command for the package manager you installed Packer with or download and replace the existing binary to upgrade Packer. Refer to Install Packer for instructions.

Upgrade from 1.11

Packer 1.12's notable changes can be summarised by the following points:

1. Protocol changes for Packer/plugin communication
2. Introduction of a Directed Acyclic Graph (DAG) for Data Sources and Locals

Note: These are not breaking changes, and your templates should continue working as-is.

Protocol changes

When initialising, Packer is now able to pick between two different protocols for communication with plugins: gob (legacy), and protobuf/msgpack.

The protobuf/msgpack protocol will become the standard approach after this release, but in order to avoid breaking retro-compatibility, Packer only chooses this protocol if the following conditions are true:

1. All the plugin candidates for a build support the new protocol
2. Packer supports the protocol

This is an internal Packer change and should be completely transparent to you. However, if you start seeing errors with Packer/plugin communication like serialization errors, please report it to us so we can fix it in a later version. You can set the PACKER_FORCE_GOB environment variable to force Packer and plugins to use the legacy serialization format for communication.

Example: export PACKER_FORCE_GOB=1 will force using gob for serialization.

Data Source and Locals DAG support

In order to allow more flexibility in how you define data sources and locals, we introduced a graph-based approach to scheduling the evaluation of those components. With this change, you can use a local variable in a data source, and vice-versa. You are not bound to a strict evaluation order.

If your templates worked with older versions of Packer, this change should be transparent to you. However, if you are experiencing errors with your data sources or locals, typically errors due to unkown values, or other dependency problems, please report it to us so we can fix it in a later version. You can use the --use-sequential-evaluation command-line option can be used to revert back to the evaluation scheduler that Packer used prior to v1.12.0.

Example: packer build --use-sequential-evaluation <TEMPLATE> will force Packer to use the sequential evaluation methods.