Changelog
Behaviour changes
- Introduce a timeout while processing Event Triggers. The timeout will be determined by the lower value, 30 minutes or the response timeout associated with the Event Trigger.
Bug fixes and improvements
Server
- Add a new Prometheus metric
hasura_websocket_message_queue_time
and extend thehasura_graphql_requests_total
to also include subscriptions requests. (Cloud / Enterprise edition only) - Add a new Prometheus metric
hasura_active_subscription_pollers_in_error_state
for better observability in the subscriptions subsystem. (Cloud / Enterprise edition only) - Add support for
pg_get_event_logs
,pg_get_event_invocation_logs
,pg_get_event_by_id
,mssql_get_event_logs
,mssql_get_event_invocation_logs
andmssql_get_event_by_id
APIs. See the docs for more details. - Throw
source-cleanup-failed
warnings when an inconsistent source is removed using thereplace_metadata
orclear_metadata
metadata APIs only if there are event triggers associated with the removed source. - Mark Event Triggers as inconsistent if GraphQL Engine is unable to create the Event Trigger infrastructure (SQL triggers and tables for logging) on the source.
- In the OpenTelemetry exporter, set the
service.name
resource attribute to "hasura" by default. (Enterprise edition only) - Improve tracing spans and OpenTelemetry exporter. The top level trace is annotated with
request_id
andparameterized_query_hash
. Bulk queries should output a separateparameterized_query_hash
for each query span. (Enterprise edition only)
Console
- Fix a UI bug in inspect modal in the operation tab under monitoring. The modal was broken if any error was associated with the inspected operation. (Cloud / Enterprise edition only)
- Updated create REST endpoint form to fix the base URL from a fixed label of
localhost:8080
to the actual server URL. - Add warning badge for non-cached queries with warning response header in API Explorer.
- The "Track All" action for tables in Data Connector-backed databases does not fail for the entire batch if there are conflicts with existing schema for a one/more tables. The ones that can be tracked will be tracked successfully.
- Clicking the name of a tracked table in the new table tracking UI will now link to the manage table page.
- Add Console support for warning notifications while importing and resetting Metadata, and setting API time limits.