Tally

Appearance

How-to guides ​

A how-to guide is a recipe. It gives the shortest correct path to one goal for a reader who already knows what they want and only needs the steps in the right order.

A guide does not teach you what the parts are; that is the job of a tutorial. It opens with what it assumes is already running and what you need at hand, and it closes with how to check that the result is what you wanted. Between those two there are steps and nothing else.

The sidebar lists the guides by area, so the tasks that belong to one part of Tally sit together.

Guides by area ​

Cluster ​

  • Deploy the stack to a cluster puts the Reporting API, the OTLP endpoint, Grafana and the OpenStack collector on a dedicated cluster behind Let's Encrypt certificates, with the secrets kept out of the repository. The cluster migrates both databases, rates every month and reconciles the cloud by itself.
  • Use another Gateway API implementation deploys the prod overlay behind Traefik or another Gateway API implementation, without the component that holds the Envoy Gateway objects.

OpenStack provider ​

  • Install the collector from the Debian package puts one collector on a control node as a systemd service, with its credentials in files only it may read, and upgrades, removes or purges it again.
  • Connect the collector to an OpenStack cloud points one collector at one cloud, so its services publish the notifications the collector reads and that cloud's events arrive at the Reporting API.
  • Issue and revoke credentials issues the ingest credential a collector reports under and the query token a reader holds, and revokes either one.
  • Verify a deployment with the dump prints what one deployment publishes, before a collector is pointed at it, and compares that against the notification mapping.
  • Reconcile a cloud gives the Reporting API the clouds file, the account and the entry a sync of one cloud needs, and runs that sync.
  • Zero-rate a service project registers the project an OpenStack service keeps its own resources in, and bills it at zero or under a project of the operator.

Engine ​

  • Migrate both databases brings the reporting and the engine schema to what their binaries expect, and rolls one chain back again.
  • Import a pricing model loads one pricing model file into the engine database, where a run resolves the prices of the period it rates.
  • Run a period meters and rates one billing month into the run that finalization, correction and export work on.
  • Finalize a period closes one billing month, so its records turn immutable and the period stops taking new ones.
  • Detect late events reports the events that arrived after the run that bills a month read it.
  • Correct a finalized period meters a finalized month again, stores every non-zero difference as a delta and renders one credit note per affected project.
  • Export a run writes one run's result into a directory, once as JSON documents and once as a CSV table.
  • Model a customer group with a discount registers a meta-project, relates the customer's projects to it and puts the group discount on those relations.
  • Export a rollup per customer group sums one run's statements under the groups its projects belong to and writes one document per group beside the statements.

Observability ​

  • Publish metrics over OTLP puts one cloud's Ceilometer samples into the metrics store on the push path, with the labels and the credential the receivers ask for.
  • Scrape the OpenStack exporters points the store's two OpenStack scrape jobs at the exporters of one deployment, and takes Ceilometer to the store through a Pushgateway.
  • Install the Grafana dashboards publishes Grafana and the four provisioned dashboards on a cluster, and checks that the hostname answers and the dashboards render.
  • Fill the dashboards on a dev cluster pushes real events and synthetic exporter series into a dev cluster until every panel of all four dashboards carries a value.
  • Deploy alerting and route notifications brings up vmalert and Alertmanager with a deployment's own external URLs, and names the receiver an alert is handed to.

Simulator ​

  • Run a simulated month against the dev cluster publishes one generated OpenStack month onto the broker the dev cluster's collector consumes from, paces it, and reads the month back through the Reporting API.
  • Replay a recorded month writes a month to files with run --out and puts one of those files back on a broker, without the generator or the seed the month came from.
  • Compare an export against the oracle holds the CSV export of a metered simulated month against oracle.json, the generator's own statement of that month.
  • Switch on fault switches runs a simulated month with fault switches on and reads what each one changes on the bus, in the counters and in the comparison.
  • Reconcile the simulated cloud syncs the dev cluster's Reporting API against the OpenStack API a running month serves, at the instant the month stands at.
  • Register simulated projects registers the tenants and the Gardener projects of a simulated month with the dev registry, and bills the month through the relations it wrote.

Respond to an alert ​

  • TallyCloudEventsSilent fires when a cloud that ingested collector events at some point in the last 24 hours has ingested none for more than an hour.
  • TallySyncStale fires when no reconciliation run of a cloud reached completed in the last 30 minutes.
  • TallyScrapeTargetDown fires when a configured scrape target has answered nothing for five minutes.
  • TallyScrapeJobMissing fires when a discovered scrape job resolves to no target at all, so the job leaves the target page instead of turning red.
  • TallyExporterServiceSilent fires when the database exporter target is up while one of the five billed services emits no series for a cloud.