Metrics#

A metric is simply put a data point on a measured entity. Where the entity could be anything that can be polled or in itself gives an output. So something you can measure can be turned into a metric. To be more specific a metric has a value, a name and possibly some annotation tags to go along with it. Metrics are always reported along with a timestamp. Making it possible to measure an entity over time. Every metric is stored in a database. A so called time-series database. A time-series database system, is optimized for storing time-series data which is exactly what a metric is.

Specifics on metrics in regards to HealOps#

• Metrics are generated whenever a Tests, Repairs or Stats file is executed.

Guidelines#

When generating metrics with HealOps ensure that:

• You give the metric a name
• That the metric has a value and the type of that value is int32
• Add some tags to a metric. At least one. Adding tags makes it possible to slice and dice (drill down into) on the metrics data in the time-series database.
• Do not be overly specific when naming a metric. A metric named e.g. mssqlserver.mysqlsrv1.cpu.1 is not a good way to go. Most time-series databases will have to store a data point for each reported metric. Resulting in extra use of storage and even more inconvenient, constraints on your drilling and dicing options on data. > Read more here < the "Naming Schema" section.