node_exporter msg=”error encoding and sending metric family: write tcp X.X.X.X:9100->Y.Y.Y.Y:53638: write: broken pipe\n” source=”log.go:172″
Prometheus 的 node_exporter 在 syslog 會出現 連續 錯誤訊息
May 28 21:41:19 pve-1 node_exporter[1409]: time=”2019-05-28T21:41:19+08:00″ level=error msg=”error encoding and sending metric family: write tcp X.X.X.X:9100->Y.Y.Y.Y:53638: write: broken pipe\n” source=”log.go:172″
May 28 21:41:19 pve-1 node_exporter[1409]: time=”2019-05-28T21:41:19+08:00″ level=error msg=”error encoding and sending metric family: write tcp X.X.X.X:9100->Y.Y.Y.Y:53638: write: broken pipe\n” source=”log.go:172″
May 28 21:41:19 pve-1 node_exporter[1409]: time=”2019-05-28T21:41:19+08:00″ level=error msg=”error encoding and sending metric family: write tcp X.X.X.X:9100->Y.Y.Y.Y:53638: write: broken pipe\n” source=”log.go:172″
May 28 21:41:19 pve-1 node_exporter[1409]: time=”2019-05-28T21:41:19+08:00″ level=error msg=”error encoding and sending metric family: write tcp X.X.X.X:9100->Y.Y.Y.Y:53638: write: broken pipe\n” source=”log.go:172″
May 28 21:41:19 pve-1 node_exporter[1409]: time=”2019-05-28T21:41:19+08:00″ level=error msg=”error encoding and sending metric family: write tcp X.X.X.X:9100->Y.Y.Y.Y:53638: write: broken pipe\n” source=”log.go:172″
May 28 21:41:19 pve-1 node_exporter[1409]: 2019/05/28 21:41:19 http: multiple response.WriteHeader calls
這是因為調整 Prometheus server 的 job 設定，將 scrape_interval 與 scrape_timeout 兩個參數都設到 1s 後開始出現，看起來是 job 抓 metrics 的作業來不及跑完又執行所造成。改為 5s 後就沒出現了。