Metrics are extracted from several sources:
In addition to the common common_message_format, metric messages have additional properties.
Attributes in bold are always present in the messages while attributes in italic are optional.
heka.sandbox.metric(for metrics derived from other messages).
- name (string), the name of the metric. See metric_list for the current metrics names that are emitted.
- value (number), the value associated to the metric.
- type (string), the metric's type, either
gauge(a value that can go up or down),
counter(an always increasing value) or
derive(a per-second rate).
- source (string), the source from where the metric comes from, it can be the name of the collectd plugin,
<service>-apifor HTTP response metrics.
- hostname (string), the name of the host to which the metric applies. It may be different from the
Hostnamevalue. For instance when the metric is extracted from an OpenStack notification,
Hostnameis the host that captured the notification and
Fields[hostname]is the host that emitted the notification.
- interval (number), the interval at which the metric is emitted (for the
- tenant_id (string), the UUID of the OpenStack tenant to which the metric applies.
- user_id (string), the UUID of the OpenStack user to which the metric applies.
Metric messages may include additional fields to specify the scope of the measurement. When this is the case, these fields are detailed in the list of metrics presented hereafter.
This is the list of metrics that are emitted by the LMA collector service. They are listed by category then by metric name.