- Add the operations/pxc-add-metrics.yml ops file to your pxc deployment.
- Change the operations/loggregator_vars_template.yml to have the correct name of your director and cf deployment, so that the loggregator agent gets the cert from the loggregator deployment in cf, and add this as a vars file.
- Set the
loggregator_agent_deployment
variable for the loggregator agent job to tag your metrics with the deployment name.
In order to use cf-mysql-release as the internal database for Cloud Foundry, various components of Cloud Foundry like UAA and CAPI require the backing database to be online before they will start successfully. However, the replication canary also has a dependency on UAA to obtain a token for sending emails via the notifications service.
The easiest way to break this dependency cycle when adding this monitoring to a Cloud Foundry deployment is as follows:
- ensure
max_in_flight=1
(mysql always needs this anyway) - ensure
serial: true
in the update block - re-order jobs in manifest as follows:
- consul
- other jobs e.g. NATS, router
- mysql DBs
- mysql proxies
- UAA/CC
- replication-canary
- everything else including Diego