How should we report missing metrics

See original GitHub issue

If you have a time series (example blob storage exporting metric transactions) you will receive a set of results as follows:

"timeseries": [ { "metadatavalues": [], "data": [ { "timeStamp": "2019-05-09T15:34:00Z" }, { "timeStamp": "2019-05-09T15:35:00Z" }, { "timeStamp": "2019-05-09T15:36:00Z" }, { "timeStamp": "2019-05-09T15:37:00Z" }, { "timeStamp": "2019-05-09T15:38:00Z" }, { "timeStamp": "2019-05-09T15:39:00Z" }, { "timeStamp": "2019-05-09T15:40:00Z" }, { "timeStamp": "2019-05-09T15:41:00Z" }, { "timeStamp": "2019-05-09T15:42:00Z" }, { "timeStamp": "2019-05-09T15:43:00Z", "total": 37 }, { "timeStamp": "2019-05-09T15:44:00Z" }, { "timeStamp": "2019-05-09T15:45:00Z" }, { "timeStamp": "2019-05-09T15:46:00Z" }, { "timeStamp": "2019-05-09T15:47:00Z" }, { "timeStamp": "2019-05-09T15:48:00Z" }, { "timeStamp": "2019-05-09T15:49:00Z" }, { "timeStamp": "2019-05-09T15:50:00Z" }, { "timeStamp": "2019-05-09T15:51:00Z" }, { "timeStamp": "2019-05-09T15:52:00Z" }, { "timeStamp": "2019-05-09T15:53:00Z" }, { "timeStamp": "2019-05-09T15:54:00Z" }, { "timeStamp": "2019-05-09T15:55:00Z" }, { "timeStamp": "2019-05-09T15:56:00Z" }, { "timeStamp": "2019-05-09T15:57:00Z" }, { "timeStamp": "2019-05-09T15:58:00Z" }, { "timeStamp": "2019-05-09T15:59:00Z" }, { "timeStamp": "2019-05-09T16:00:00Z" }, { "timeStamp": "2019-05-09T16:01:00Z" }, { "timeStamp": "2019-05-09T16:02:00Z" }, { "timeStamp": "2019-05-09T16:03:00Z" }, { "timeStamp": "2019-05-09T16:04:00Z" }, { "timeStamp": "2019-05-09T16:05:00Z" }, { "timeStamp": "2019-05-09T16:06:00Z" }, { "timeStamp": "2019-05-09T16:07:00Z" }, { "timeStamp": "2019-05-09T16:08:00Z" }, { "timeStamp": "2019-05-09T16:09:00Z" }, { "timeStamp": "2019-05-09T16:10:00Z" }, { "timeStamp": "2019-05-09T16:11:00Z" }, { "timeStamp": "2019-05-09T16:12:00Z" }, { "timeStamp": "2019-05-09T16:13:00Z" }, { "timeStamp": "2019-05-09T16:14:00Z" }, { "timeStamp": "2019-05-09T16:15:00Z" }, { "timeStamp": "2019-05-09T16:16:00Z" }, { "timeStamp": "2019-05-09T16:17:00Z" }, { "timeStamp": "2019-05-09T16:18:00Z" }, { "timeStamp": "2019-05-09T16:19:00Z" }, { "timeStamp": "2019-05-09T16:20:00Z" }, { "timeStamp": "2019-05-09T16:21:00Z" }, { "timeStamp": "2019-05-09T16:22:00Z" }, { "timeStamp": "2019-05-09T16:23:00Z" }, { "timeStamp": "2019-05-09T16:24:00Z" }, { "timeStamp": "2019-05-09T16:25:00Z" }, { "timeStamp": "2019-05-09T16:26:00Z" }, { "timeStamp": "2019-05-09T16:27:00Z" }, { "timeStamp": "2019-05-09T16:28:00Z" }, { "timeStamp": "2019-05-09T16:29:00Z" }, { "timeStamp": "2019-05-09T16:30:00Z" }, { "timeStamp": "2019-05-09T16:31:00Z" }, { "timeStamp": "2019-05-09T16:32:00Z" }, { "timeStamp": "2019-05-09T16:33:00Z" } ] } ]

However when promitor queries this it returns -1 for any query that doesn’t return a value.

I believe it makes more sense to return 0 in this case as advised by prometheus “Avoid missing metrics” https://prometheus.io/docs/practices/instrumentation/#counter-vs-gauge-summary-vs-histogram