Status: Closed (View Workflow)
Currently, if mod-search is not initialized for the tenant, the error will be thrown, which affects message processing for other tenants.
Also current retry settings for failed messages are:
application: kafka: retry-interval-ms: 20000 retry-delivery-attempts: 10
This leads that initial bulk processed with a time equal to:
10 * 20 sec = 200 sec
and then, if it's failing every message is processed with the same amount of time. So the gracefully handled bulk processed with time:
200s + 50 (amount of messages in bulk) * 10 * 20 sec = 10200 sec = 170 min
This value must be decreased to 10 minutes to not stop the processing of messages for a long time. The current implementation will log every unprocessed message.
- If mod-search is not initialized for the tenant, the error should not be thrown.
- Kafka settings: retry-interval-ms and retry-delivery-attempts must be decreased
- retry-interval-ms must be set to 2000
- retry-delivery-attempts must be set to 6
- Make these options configurable by setting environment variables: KAFKA_RETRY_INTERVAL_MS and KAFKA_RETRY_DELIVERY_ATTEMPTS
- Integration and unit tests are passed through
UXPROD-3046 NFR: Inventory. Swap from search tool PostgreSQL to Elasticsearch
UXPROD-3271 NFR: Look up instance plug-in. Swap from search tool PostgreSQL to Elasticsearch
UXPROD-3430 Inventory. Implement Facets on filtering (Instance, Holdings, Item) when using Elasticsearch