Setting the file. One moment.
Subchapter 42.1
enums.mdMarkdown2 KBView on GitHub
Load this file when you’re about to call create-graph or preview-graph. The API is strict about casing — wrong case means the call is rejected.
productType (PascalCase, required)Use exactly one of:
LogsTracesErrorsSessionsMetricsEventsDo NOT use sessions, errors, logs, traces, metrics (lowercase). The API rejects them.
Each expression MUST have both keys:
{
"aggregator": "Count",
"column": ""
}CountCountDistinctCountDistinctKeyMinAvgP50P90P95P99MaxSumDo NOT use count_distinct, Count_distinct, countdistinct, or any snake_case variant. It’s CountDistinct — PascalCase, no underscore.
Even for Count where no column is needed. Pass an empty string:
{"aggregator": "Count", "column": ""}Omitting column crashes the call with an argument-validation error.
Line chartBar chartTableDefault: Timestamp. Usually fine to omit or leave as default. For non-time aggregations (e.g., grouping by service across a fixed window), omit bucketBy and rely on groupBy.
Different from create-graph — get-keys wants lowercase productType and full ISO timestamps:
productType: lowercase — logs, traces, sessions, errors, metrics, eventsstartDate / endDate: full ISO with time component — 2026-03-04T00:00:00Z, not date-only 2026-03-04Yes, create-graph wants Sessions (PascalCase) and get-keys wants sessions (lowercase). It’s inconsistent. Just remember: get-keys is lowercase, the graph tools are PascalCase.
Before calling create-graph, re-read your arguments:
projectKey present? ✓productType PascalCase? ✓expressions is a list of objects each with aggregator and column? ✓aggregator is PascalCase (no underscore)? ✓column present even for Count? ✓If any answer is no, fix it before calling. API errors are expensive round-trips.