Metrics
Open Object API produces application metrics (using Open Telemetry).
Note
The exact metric names that show up may be transformed, e.g. Prometheus replaces periods with underscores, and processing pipelines may add prefixes or suffixes.
Important
Some metrics are defined as “global scope”.
These metrics are typically derived from application state introspection, e.g. by performing database (read) queries to aggregate some information. Usually those correspond to an Asynchronous Gauge.
Multiple replicas and/or instances of the same service will produce the same values
of the metrics. You need to apply some kind of aggregation to de-duplicate these
values. The attribute scope="global" acts as a marker for these type of metrics.
With PromQL for example, you can use avg on the assumption that all values will
be equal, so the average will also be identical:
avg by (type) (otel_openobject_auth_user_count{scope="global"})
Generic
http.server.durationCaptures how long each HTTP request took, in ms. The metric produces histogram data.
http.server.request.duration(not active)The future replacement of
http.server.duration, in seconds. Currently not enabled, but the code is in the Open Telemetry SDK instrumentation already and could possibly be opted-in to.
Application specific
Accounts
openobject.auth.user_countReports the number of users in the database. This is a global metric, you must take care in de-duplicating results. Additional attributes are:
scope- fixed, set toglobalto enable de-duplication.type- the user type.all,stafforsuperuser.
Sample PromQL query:
max by (type) (last_over_time( otel_openobject_auth_user_count{scope="global"} [1m] ))
openobject.auth.login_failuresA counter incremented every time a user login fails (typically because of invalid credentials). Does not include the second factor, if enabled. Additional attributes:
http_target- the request path where the login failure occurred, if this happened in a request context.
openobject.auth.user_lockoutsA counter incremented every time a user is locked out because they reached the maximum number of failed attempts. Additional attributes:
http_target- the request path where the login failure occurred, if this happened in a request context.username- username of the user trying to log in.
openobject.auth.loginsCounter incrementing on every successful login by a user. Additional attributes:
http_target- the request path where the login failure occurred, if this happened in a request context.username- username of the user trying to log in.
openobject.auth.logoutsCounter incrementing every time a user logs out. Additional attributes:
username- username of the user who logged out.
Objects
openobject.object.createsReports the number of objects created via the API.
openobject.object.updatesReports the number of objects updated via the API.
openobject.object.deletesReports the number of objects deleted via the API.
Objecttypes
openobject.objecttype.createsReports the number of objecttypes created via the API.
openobject.objecttype.updatesReports the number of objecttypes updated via the API.
openobject.objecttype.deletesReports the number of objecttypes deleted via the API.
The metrics show how many entities are created, updated, or deleted via the API, helping to monitor load and the most frequent operations, and allow for various aggregations on the data.
Sample PromQL query:
sum by (otel_scope_name) (otel_openobject_object_updates_total)