Near No-Downtime Upgrades: Discovery on dsva-vagov-prod

jennb33 commented 3 months ago

Problem

In order to reduce or even remove the need for downtime during future upgrades of this cluster, we need to do discovery on whether or not the clusters will rebuild themselves if there is data loss. If this cache simply consists of things that the vagov framework caches, then we should be able to forego backups and take advantage of zero/near-zero downtime upgrades of these clusters.

Questions to be answered

[ ] Given that we have one cluster, but many pods, how do the pods know which of the contents are their caches?
[ ] Given that we have one cluster, but many pods, how do the behave if the data is suddenly gone?
[ ] What changes would have to be made to make this cluster truly ephemeral (if any)?

Tasks

[ ]

Success Metrics

[ ] ADD SUCCESS METRICS HERE

Acceptability Criteria

[ ] ADD A/C HERE

va-vsp-bot commented 3 weeks ago

This issue is stale. If it is no longer valid, please close issue. Otherwise please update

jennb33 commented 3 weeks ago

This issue is not stale.

department-of-veterans-affairs / va.gov-team