This page contains the list of deprecations and important or breaking changes for Vault 1.3.0 compared to 1.2.4. Please read it carefully.
There has been a change to the way that activating performance and DR secondary clusters works when using public keys for encryption of the parameters rather than a wrapping token. This flow was experimental and never documented. It is now officially supported and documented but is not backwards compatible with older Vault releases.
On its cluster port, Vault will no longer advertise the full TLS 1.2 cipher suite list by default. Although this port is only used for Vault-to-Vault communication and would always pick a strong cipher, it could cause false flags on port scanners and other security utilities that assumed insecure ciphers were being used. The previous behavior can be achieved by setting the value of the (undocumented) cluster_cipher_suites config flag to tls12.
The API now allows multiple options for how it deals with renewals. The legacy behavior in the Agent/API is for the renewer (now called the lifetime watcher) to exit on a renew error, leading to a reauthentication. The new default behavior is for the lifetime watcher to ignore 5XX errors and simply retry as scheduled, using the existing lease duration. It is also possible, within custom code, to disable renewals entirely, which allows the lifetime watcher to simply return when it believes it is time for your code to renew or reauthenticate.
As of 1.3.0, paths that were once filtered with Filtered Mount Replication are migrated to use Filtered Paths Replication. The APIs are very similar, but as of 1.3.0, we have marked the mount filter API as deprecated and recommend migrating to the path filter API, as the mount filter API will be removed in a future version of Vault.
In versions of Vault prior to 1.3.0, mount filters could be used to allow only specified mounts to replicate to secondary nodes via an allowlist (referred to as a whitelist in the original documentation of the feature). After upgrade to 1.3.0 and later, if allowlist/whitelist mount filters were enabled on a secondary node, namespaces and their default mounts will also stop replicating to that secondary node.