Hello, this is 
id:kmuto from the Mackerel CRE team. As previously announced, development is well underway for the beta release of Mackerel's log feature. However, we certainly haven't forgotten about our daily feature improvements to continue delivering value to our users! Here are the details of this week's updates.
- Series selection now available for custom dashboard graphs
- Azure Cosmos DB metrics can now be retrieved via Azure Integration
- Banner messages during maintenance and incidents now indicate the affected functional scope
- New PromQL syntax format now supported for specifying metrics
- The use_plugin_timestamp parameter is now available for custom metric submissions
Series selection now available for custom dashboard graphs
In Mackerel, a single graph area may display multiple data series. Examples include graphs showing memory usage and free space, CPU utilization per Docker container, or HTTP code counts per ALB target.
While you could click to select which graph series to display on the host screen, the graph widgets placed on custom dashboards displayed all series without an option to select them.
With this update, you can now select the series to display for host graphs and service metric graphs within the custom dashboard graph widget settings.
You can easily switch the view to show only the series you want to focus on, making dashboard monitoring much more efficient!
In addition, the dashboard creation, retrieval, and update APIs also support configuring and retrieving this graph series selection via the legendList key of the graph widget object.
Configuration is also supported in Terraform provider for Mackerel via the legend_list argument (since v0.12.0).
Azure Cosmos DB metrics can now be retrieved via Azure Integration
Azure Cosmos DB is a fully managed NoSQL database service provided by Azure. Metrics for this service can now be retrieved using Mackerel's Azure Integration.
Each database account is counted as 1 micro host. For more details, please see the help documentation below.
Banner messages during maintenance and incidents now indicate the affected functional scope
During maintenance and incidents, Mackerel provides detailed information on our status page (https://status.mackerel.io). We also display a banner at the top of the Mackerel Web Console to notify users that maintenance or an incident is ongoing.
However, previously, these banners only stated "Currently, some systems of Mackerel are under maintenance." or "Currently, some systems of Mackerel are experiencing trouble." To check which specific components were affected and how they impacted the features you use, you had to visit the status page and read the descriptions.
With this improvement, the banner now includes the specific features being affected. While you will still need to check the status page for detailed progress and updates, this will help you quickly determine whether the features you are using are impacted.
New PromQL syntax format now supported for specifying metrics
In Mackerel, PromQL syntax is used when creating query monitors or query graphs for labeled metrics. Alongside the metric_name{"label_name"="value"} syntax for specifying metrics with labels, the newly introduced PromQL syntax format {"metric_name", "label_name"="value"} is now supported.
Additionally, quotation mark auto-completion and label suggestions are provided when using this new syntax.
The use_plugin_timestamp parameter is now available for custom metric submissions
When metrics are collected and submitted using a mackerel-agent plugin, mackerel-agent sets the submission timestamp based on the time it collected the data from the plugin. However, if you are building a custom plugin, for example, one that retrieves metrics from a cloud provider from a slightly earlier time, you might want to explicitly specify a particular timestamp rather than using the current collection time.
To support this use case, mackerel-agent v0.87.0 introduces the use_plugin_timestamp parameter. This allows you to configure, on a per-plugin basis, whether to use the exact timestamp provided by the plugin without overwriting it.
At present, official plugins are designed to use the execution date and time as the metric submission timestamp, so you won't notice any change with them under this setting. However, if you are using custom plugins, this will allow you to correct any timestamp discrepancies.
For more details, please check the documentation below.
