-
Notifications
You must be signed in to change notification settings - Fork 54
Fix classification when adding some components #258
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
a6806a7
to
9ea0cf5
Compare
9c9945c
to
9085054
Compare
f6f5b7e
to
828050f
Compare
Fixes the lack of classificaton in the add_replica plan so that it does not fail when adding a replica to a deployment which was not previously configured with one. Without this fix, the plan could only replace failed replicas of the same name.
Changes to add_replica which fix classificaton invalidate tests, commit makes them valid again.
Previous to this, utility plan update_classification made unnecessary assumptions about primaries and replicas. This commit ensure those assumptions are not made and classification is based solely on availability group letter.
The switch to availability group based classification necessitated changes to add_database for it to continue working. Does a little clean up of various cruft along the way.
It is not guaranteed to be in path
When reusing failed infrastructure components they may be configured for a different primary then is current and have an old certificate revacation list. Commit ensures that agent configuration is updated for the current primary and fetches CRL from that primary. Includes a little cleanup lifted from the add_compiler plan.
When running peadm::subplans::modify_certificate also get status of certificate from the perspective of the primary to detect if the certificate has been revoked. Introduces new task, peadm::cert_valid_status which checks different failure scenarios when validating certificates.
Acceptable failures when running clean on a primary expanded to address scenarios where an infrastructure component is cleaned by another process, e.g. puppet infrastructure forget
Creates a utility plan that is used by add_replica plan to source the primary's global hiera configuration and distribute it to replicat target. Without this, data in the console is not available when compiling catalogs after replica is promoted.
Capability to set PuppetDB database backend address to anything. Previously, peadm::util::update_db_setting would always attempt to pair configuration with appropriate availability group letter but in DR scenarios this is not appropriate.
The addition of the peadm::cert_valid_status task triggered test suite failures. Commit fixes them.
Ran through orchestrator. Bolt runs successfully until reaching the final step when it reported that it failed to connect to the rbac-api.
Orchestrator does continue running the final task though and it is ultimately successful, resulting in a functional fully provisioned Replica. Test was completed on the CLI via Bolt, using a token. I presume this is related to the |
Noticed today while testing some workflows that the add_database code does not take into consideration postgres 14 |
Hey @ody the logic looks good to me 👍 |
Ensure classification is updated appropriately. Lacking classification results in plans only being able to replace components if they are of the same name but entirely unconfigured.