GetClusterHealthWaitForAllEvents may return empty/default value

[barkbay]

If the Elasticsearch endpoint "/_cluster/health?wait_for_events=languid&timeout=0s" returns 408 (Timeout) then GetClusterHealthWaitForAllEvents swallows the error and returns an empty/default value. TimedOut may hold the default value false while the API actually returned true:

{
	"cluster_name": "elasticsearch-sample",
	"status": "green",
	"timed_out": true,
	"number_of_nodes": 3,
	"number_of_data_nodes": 3,
	"active_primary_shards": 12,
	"active_shards": 24,
	"relocating_shards": 0,
	"initializing_shards": 0,
	"unassigned_shards": 0,
	"delayed_unassigned_shards": 0,
	"number_of_pending_tasks": 0,
	"number_of_in_flight_fetch": 0,
	"task_max_waiting_in_queue_millis": 0,
	"active_shards_percent_as_number": 100.0
}

The returned value is not initialized because we do not attempt to parse/decode the body and return immediately when there is an error:

func (c *baseClient) request(
	...
	responseObj interface{},
) error {
	resp, err := c.doRequest(ctx, request)
	if err != nil {
		return err // <--- here
	}
    ...
	if responseObj != nil {
		if err := json.NewDecoder(resp.Body).Decode(responseObj); err != nil {
			return err
		}
	}

Should we attempt to decode the body as a best effort even if there is an error ?

[barkbay]

408 should be replaced by 200 in the future: elastic/elasticsearch#70849 (comment)