Files
vault/changelog/22235.txt
Alexander Scheel be2f109275 Ignore errors from rollback manager invocations (#22235)
* Ignore errors from rollback manager invocations

During reload and mount move operations, we want to ensure that errors
created by the final Rollback are not fatal (which risk failing
replication in Enterprise when the core/mounts table gets invalidated).
This mirrors the behavior of the periodic rollback manager, which
only logs the error.

This updates the noop backend to allow failing just rollback operations,
which we can use in tests to verify this behavior and ensure the core
operations (plugin reload, plugin move, and seal/unseal) are not broken
by this. Note that most of these operations were asynchronous from the
client's PoV and thus did not fail anyways prior to this change.

Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>

* Add changelog entry

Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>

* Update vault/external_tests/router/router_ext_test.go

Co-authored-by: Nick Cabatoff <ncabatoff@hashicorp.com>

---------

Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
Co-authored-by: Nick Cabatoff <ncabatoff@hashicorp.com>
2023-08-08 15:42:02 +00:00

4 lines
147 B
Plaintext

```release-note:improvement
core: Log rollback manager failures during unmount, remount to prevent replication failures on secondary clusters.
```