Release notes 3.8.12

Edit this page

This is a bugfix release, no new features since 3.8.11 were added.

Changelog

Compared to the latest release 3.8.11, the following changes and bugfixes were added.

Bugfixes

This is a complete list of the merged pull requests.

Fix background sync always detecting policy inconsistency for Things with policy imports

PR #2317 fixes an issue where background sync always detected a policy inconsistency for Things whose Policies have imports. The MongoDB projection in sudoStreamMetadata() was missing the referenced policies field, causing the indexed metadata to always have an empty set of referenced policies. This triggered continuous unnecessary search index updates.

Fix aggregate metrics using wrong scrapeInterval and showing 0ms duration

PR #2334 fixes two bugs in operator aggregate metrics:

• OperatorAggregateMetricsProviderActor used the maximum configured scrapeInterval for all metric timers instead of each metric’s own interval. Now each metric’s individual scrapeInterval is respected, falling back to the global default.
• AggregateThingsMetricsActor measured aggregation duration at stream materialization time instead of at stream completion time, causing the logged duration to always be ~0ms.

Fix Policy activation failure when other expired subjects exist

PR #2335 fixes a bug where activateTokenIntegration requests failed when the Policy contained other subjects that were already expired but still within their grace period before deletion. Now only the requested subject is checked for being expired, instead of checking the entire Policy.

Helm Chart

Expose Pekko HTTP host pool max connection lifetime via Helm values

PR #2336 adds connectivity.config.connections.httpPush.maxConnectionLifetime to Helm values (default: "Inf") and wires it to the PEKKO_HTTP_HOSTPOOL_MAX_CONNECTION_LIFETIME environment variable in the connectivity deployment template.