Aggregation shows 0 instead of actual value


(Mikael Selander) #1

Hey,

I'm having a mysterious issue with my Kibana visualizations. For some reason the CPU and memory graphs flatline to 0 after a specific date. The data is there, though. Searching for it in Discover shows non-0 values for both metrics. The data is sent from metricbeats, creating one index per day. The memory graph flatlined over 1 day and then came back, the CPU graph has been at 0 for the last couple of days.

These are the graphs:
CPU

Memory
Memory

CPU data in Discover

Query generated by the Kibana Visualization
{"index":["metrics-*"],"ignore_unavailable":true,"preference":1536218298985} {"size":0,"_source":{"excludes":[]},"aggs":{"2":{"date_histogram":{"field":"@timestamp","interval":"3h","time_zone":"Europe/Berlin","min_doc_count":1},"aggs":{"3":{"terms":{"field":"beat.hostname.keyword","size":5,"order":{"1":"desc"}},"aggs":{"1":{"max":{"field":"system.process.cpu.total.norm.pct"}}}}}}},"stored_fields":["*"],"script_fields":{},"docvalue_fields":["@timestamp","system.process.cpu.start_time"],"query":{"bool":{"must":[{"query_string":{"query":"metricset.name:process AND system.process.cmdline:\"/usr/bin/dotnet /opt/upc/bin/Mg.Public.RequestResponse.dll\"","analyze_wildcard":true,"default_field":"*"}},{"range":{"@timestamp":{"gte":1535787966153,"lte":1536219966153,"format":"epoch_millis"}}}],"filter":[],"should":[],"must_not":[]}}}

Query response (truncated)
{ "key_as_string": "2018-09-03T21:00:00.000+02:00", "key": 1536001200000, "doc_count": 2160, "3": { "doc_count_error_upper_bound": 0, "sum_other_doc_count": 0, "buckets": [ { "key": "ip-10-153-99-177.ec2.internal", "doc_count": 1080, "1": { "value": 0.0024999999441206455 } }, { "key": "ip-10-153-99-101.ec2.internal", "doc_count": 1080, "1": { "value": 0.0020000000949949026 } } ] } }, { "key_as_string": "2018-09-04T00:00:00.000+02:00", "key": 1536012000000, "doc_count": 2160, "3": { "doc_count_error_upper_bound": 0, "sum_other_doc_count": 0, "buckets": [ { "key": "ip-10-153-99-177.ec2.internal", "doc_count": 1080, "1": { "value": 0.0024999999441206455 } }, { "key": "ip-10-153-99-101.ec2.internal", "doc_count": 1080, "1": { "value": 0.0020000000949949026 } } ] } }, { "key_as_string": "2018-09-04T03:00:00.000+02:00", "key": 1536022800000, "doc_count": 2160, "3": { "doc_count_error_upper_bound": 0, "sum_other_doc_count": 0, "buckets": [ { "key": "ip-10-153-99-101.ec2.internal", "doc_count": 1080, "1": { "value": 0.0 } }, { "key": "ip-10-153-99-177.ec2.internal", "doc_count": 1080, "1": { "value": 0.0 } } ] } }, { "key_as_string": "2018-09-04T06:00:00.000+02:00", "key": 1536033600000, "doc_count": 2160, "3": { "doc_count_error_upper_bound": 0, "sum_other_doc_count": 0, "buckets": [ { "key": "ip-10-153-99-101.ec2.internal", "doc_count": 1080, "1": { "value": 0.0 } }, { "key": "ip-10-153-99-177.ec2.internal", "doc_count": 1080, "1": { "value": 0.0 } } ] } },

Cluster status is green, and all indices report as green. It seems to me as if the aggregation fails on one or multiple indices, but I get no error details. Where do I start looking for the error? What could be the issue?

Best Regards,
Mikael Selander


(Mikael Selander) #3

This issue was caused by not having proper index templates set up. ES would apply a best-guess data type, which for some indices would cause truncation of values to 0.


(system) #4

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.