1.53. Release Notes 0.92

1.53.1. New features in this release

  • Monitor health check endpoint. The url is

    http://host:9200/api/metrics/healthcheck
    

This endpoint returns small JSON dictionary that looks like this:

{"AreaMonitor":{"healthy":true},"Summary":{"healthy":true},"hbase":{"healthy":true}}

Item “Summary” indicates overall health of the monitor. If either area monitor (the component that polls devices) or database connector fails, Summary.healthy becomes false.

  • Graphs

You can add multiple variables to the graph in the device details page. Just click numbers in the table that shows current values of the monitoring variables for interfaces or hardware components. To remove variables from the graph, click [x] next to its name in the graph legend

  • Filter by tags in the device details page

You can switch between logical operations “OR” and “AND” in the filter by tags.

  • Filter by interface or hardware component name

In addition to the filter by tags, you can filter interfaces and hardware components by name. The filter matches the name as substring (no regular expressions and no globbing), case-insensitive. Filters by tags and by name are combined with logical “AND”.

1.53.2. Documentation

Documentation on the Python functions used to process monitoring data with examples in the file doc/rules.md

1.53.3. Known issues

Data processing functions should be applied to the time series data in correct order. Some of the functions operate on the time series and return new time series, while others operate only on the last data point and return single data point. For example, rate() operates on the time series and returns new time series that is one data point shorter. smm() takes time series as an input and returns new time seties as well. Functions mul() and div() operate only on the last value and return single data point. This means, sequence

mul(smm(rate( ... )))

will compute rate, smooth it and then multiple the last value, which leads to the correct result. However,

smm(mul(rate( ... )))

does not, because mul() returns only single data point so smm() does not have enough time series data to smooth.

This is going to be fixed in one of the subsequent releases.