Support diagnostics¶

When you raise a support case, TrackMe can generate a diagnostic archive — a .tgz bundle of the configuration and state TrackMe Support needs to investigate — instead of you hand-collecting KV-store records, metrics, and logs. It is available from the TrackMe UI and the REST API, and requires the power capability (trackmepoweroperations).

The UI is reached from API & tooling → Support - generate diags. You pick the diagnostic mode, scope it (tenant / component / entities for an entity diag), optionally enable anonymisation, then Generate diagnostic archive — nothing is collected or transmitted until you click it, and the archive is streamed to your browser so no data leaves the instance until you choose to share the file:

What goes in the bundle¶

You generate the archive at one of two scopes:

Entity-scoped — a specific tenant and component, plus up to 20 entities. The bundle collects those entities’ KV records, their recent metrics (last 7 days), and their latest summary and notable events — and, since 2.4.3, their ML outlier models (the outlier rules, the fitted data, and the native ML models).
Global — the central (cross-tenant) KV collections, plus the last 24 hours of scheduler and runtime metrics.

Optionally, anonymise tenants: real tenant identifiers and aliases are replaced with anonymised tokens in both filenames and content, and the mapping is returned to you so you can still correlate. Use this when the environment is sensitive.

How it runs¶

Generation is asynchronous — you start a job, poll its status, and download the archive once it completes. Through the REST API:

# start a job — returns a job_id
| trackme mode=post url="/services/trackme/v2/support_diag/generate" body="{ ... }"

# poll the job (queued / running / complete / error / cancelled)
| trackme mode=get url="/services/trackme/v2/support_diag/status" body="{'job_id': '<id>'}"

# cancel a running job
| trackme mode=delete url="/services/trackme/v2/support_diag/cancel" body="{'job_id': '<id>'}"

When the job is complete, download the archive with the one-time token returned by the status call. Jobs run a few at a time, time out after about an hour, and their records are cleaned up shortly after completion. The exact request parameters (scope, entity list, anonymisation) are documented in the REST API Reference UI.

Tip

Attach the generated .tgz to your support case. If your data is sensitive, enable tenant anonymisation when you generate it.