At this point I am on Bitnami Helm chart version v10.4.0 (deploying Discourse version v3.0.6), and recently this same failure mode began recurring after many months of not being a problem. This recurrence was not coincident with any change in deployment configuration or update of Discourse itself.
@obeyler do you have any more experience to share on this issue? In the meantime I plan to work through updates of the Discourse Helm chart as I see that I am two versions behind (latest is v12.0.3).