Closed
Description
Logstash is crashing with OOM when documents are bigger than the Elasticsearch http.max_content_length. Those documents are filling up the memory and are never freed due to infinite retries.
it seems there’s a clear intention to handle 413 status here, which is not working properly due to the manticore's adapter logic. This adapter raises an error, making the http_client condition unreachable, leading to infinite retries and no options to DLQ those problematic events (dlq_custom_codes).
Relates to: #785