4.46.0 (2023-06-09)
Features Added
- Added the capability to filter request-level metrics based on diagnostic thresholds. Request-level metrics usually are used to capture metrics per backend endpoint/replica - a high cardinality dimension. Filtering by diagnostic thresholds reduces the overhead - but also means request-level metrics can only be used for debugging purposes - not for monitoring purposes. So, it is important to use the unfiltered operation-level metrics for health monitoring in this case. - See PR 35114
- Added optional tags/dimensions for PartitionId/ReplicaId as alternative to ServiceAddress for direct-mode (rntbd) request-level metrics. - See PR 35164
- Added request level info including timeline and system usage to the
CosmosDiagnosticsContext
. - See PR 35254 and PR 35405 - Added an optional dimension/tag
OperationSubStatusCode
for operation-level metrics. - See PR 35334 - Added support for
ComputedProperty
inCosmosContainerProperties
- See PR 35046
Breaking Changes
- Renamed the JVM configuration -
COSMOS.DEFENSIVE_WARMUP_CONCURRENCY
toCOSMOS.OPEN_CONNECTIONS_CONCURRENCY
- See PR 34859
Bugs Fixed
- Enabled connection warm-up to continue in a best-effort manner to other regions in case of address resolution errors for a particular region - See PR 35323
- Fixed an issue with
ChangeFeedProcessor
to fetch all changes before delay based on configuredPollDelay
. - See PR 35324
Other Changes
- Refactored
CosmosContainerProactiveInitConfigBuilder
to make use ofContainerDirectConnectionMetadata
and to wireDirectConnectionConfig
with
JVM configuration -COSMOS.MIN_CONNECTION_POOL_SIZE_PER_ENDPOINT
- See PR 34859 - Extending maximum retry delay in
SessionTokenMismatchRetryPolicy
. - See PR 35360