Hmm. Okay, let's figure this out.
Let's run the query to fetch the Search Rate
graph for one of these black-out periods and see what the data is telling us:
Fill in the <cluster_uuid>
with the right cluster and then adjust the time period to the affected time period.
POST .monitoring-es-*/_search
{
"query": {
"bool": {
"filter": [
{
"term": {
"cluster_uuid": "<cluster_uuid>"
}
},
{
"range": {
"timestamp": {
"gte": "2019-12-16T00:51:07.080Z",
"lte": "2019-12-16T18:51:07.080Z"
}
}
}
]
}
},
"aggs": {
"check": {
"date_histogram": {
"field": "timestamp",
"fixed_interval": "30s"
},
"aggs": {
"metric": {
"max": {
"field": "indices_stats._all.total.search.query_total"
}
},
"metric_deriv": {
"derivative": {
"buckets_path": "metric",
"gap_policy": "skip",
"unit": "1s"
}
}
}
}
}
}
Let's see if this helps us. Thanks!