*What*
Use TimescaleDB in the Metrics APB for Prometheus long term storage
*Why*
App metrics are being stored in Postgres.
App metrics is currently in the same APB as Server metrics i.e. Grafana & Prometheus.
During the App Metrics POC, Postgres (with TimescaleDB) was used to store app metrics, with a view of it also being the long term store for Prometheus data.
However, based on initial investigation getting TimescaleDB running in openshift is complex. (Docker locally was used just to prove the concept).
So, for now, only Postgres (from RHSCL) is being added to the Metrics APB for App Metrics storage.
This task is to capture the potential for re-using that same Postgres instance in the Metrics APB for long term storage of Prometheus data (see Prometheus docs for why it's not recommended to just use local storage (the default storage) for long term data).
See comment https://issues.jboss.org/browse/AEROGEAR-2190?focusedCommentId=13536829&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-13536829 for more rational
*Background*
* The official TimescaleDB docker image runs as root so is not suitable for openshift
* There is an rhscl image for postgres 9.6 which timescale requires but the installation process for timescaledb is non-trivial and would require additions to the rhscl image
* An option would be to extend the rhscl image ( https://github.com/sclorg/postgresql-container/tree/master/latest#extending-image ) but again this is non-trivial and requires investigation |
|
