Open anandgia opened 4 years ago
We see the same issue all the time. I don't think the exporter is maintained much. We always increase the scrape interval and timeout to address it (not a fantastic solution).
After increasing the timeout for prom, the message didn't appear in the log.
We are running version 0.8.0 of postgres_exporter and notice these errors sporadically:
time="2020-08-25T20:32:43Z" level=info msg="Established new database connection to \"configdb.local.:5432\"." source="postgres_exporter.go:878" time="2020-08-25T20:32:43Z" level=info msg="Semantic Version Changed on \"configdb.local.:5432\": 0.0.0 -> 10.11.0" source="postgres_exporter.go:1405" time="2020-08-25T20:32:47Z" level=info msg="Starting Server: :9187" source="postgres_exporter.go:1672" 2020/08/25 22:04:35 http: multiple response.WriteHeader calls 2020/08/25 23:01:35 http: multiple response.WriteHeader calls 2020/08/25 23:02:34 http: multiple response.WriteHeader calls 2020/08/25 23:03:34 http: multiple response.WriteHeader calls 2020/08/25 23:04:34 http: multiple response.WriteHeader calls 2020/08/25 23:07:34 http: multiple response.WriteHeader calls 2020/08/25 23:09:34 http: multiple response.WriteHeader calls 2020/08/26 00:02:34 http: multiple response.WriteHeader calls 2020/08/26 00:03:36 http: multiple response.WriteHeader calls 2020/08/26 00:04:34 http: multiple response.WriteHeader calls 2020/08/26 00:05:35 http: multiple response.WriteHeader calls 2020/08/26 00:06:36 http: multiple response.WriteHeader calls 2020/08/26 00:07:44 http: multiple response.WriteHeader calls 2020/08/26 00:11:08 http: multiple response.WriteHeader calls 2020/08/26 00:11:14 http: multiple response.WriteHeader calls 2020/08/26 00:11:35 http: multiple response.WriteHeader calls 2020/08/26 00:11:44 http: multiple response.WriteHeader calls 2020/08/26 01:01:34 http: multiple response.WriteHeader calls 2020/08/26 01:08:34 http: multiple response.WriteHeader calls 2020/08/26 01:09:34 http: multiple response.WriteHeader calls 2020/08/26 03:02:34 http: multiple response.WriteHeader calls
It appears the scrape timeout expires and multiple calls at same time causes this race condition. This issue seems to be have been addressed in later version of prometheus client https://github.com/prometheus/graphite_exporter/issues/123 but we notice postgres_exporter is using an older client library.