Announcements

#mackerelio

P90, P95, and P99 latency graphs will now be displayed on the summary screen, and more

Hello, this is Mackerel team CRE's id:KGA . I would like to inform you about recent updates.

We’re improving our APM feature

We’re continuing to improve Mackerel’s APM feature, which was officially released on May 1st.

P90, P95, and P99 latency graphs will now be displayed on the summary screen

Graphs for the 90th percentile (P90), 95th percentile (P95), and 99th percentile (P99) of latency will now be displayed on the summary screen. This will make it easier to understand latency trends excluding outliers.

You can now drag to zoom in on summary graphs

You can now zoom in on graphs on the APM summary screen by dragging across the graphs. This enables you to quickly check detailed data on a specific time period.

You can now check the total number of trace spans, daily usage, and trends in overall usage in graphs. Use this capability to estimate usage fees, monitor variation in span counts, and adjust sampling.

Data retention period is now two weeks

During the beta, span data was stored for three days, but with the official release, the data retention period has been extended to two weeks.

The “HTTP Server” and “Database” pages in the side menu on the trace screen have been integrated into tabs on the APM screen

The “HTTP Server” and “Database” pages in the side menu on the trace screen have been integrated into tabs on the APM screen, enabling you to quickly check the performance of the HTTP server and database of the service displayed in APM.

AWS Integration now supports Amazon ElastiCache for Valkey

AWS Integration now supports Amazon ElastiCache for Valkey. Please see the following documentation for information on the available metrics and how to configure AWS Integration.

Added monitoring rule names to PagerDuty notifications

We added monitor rule names to the notifications sent to notification channels configured to use the Events API v2 in PagerDuty. The monitoring rule name will be included in the event details under the key monitor_name. This will make it easier to check which monitoring rule caused the notification when a PagerDuty notification is received.

{
 "client": "Mackerel",
 "client_url": "https://mackerel.io/orgs/Hatena-org/alerts/5ryhYHNjTSQ",
 "contexts": null,
 "description": "[Hatena-org] CRITICAL: https://hatena.blog/404 HTTP request failed - status is 404",
 "event_type": "trigger",
 "incident_key": "alert-XXXXXXXXXXXXXXXXXXXXXXXXXX",
 "service_key": "XXXXXXXXXXXXXXXXXXXXXXXXXXXX",
 "details": {
   "alert_created_at": 1744940159,
   "alert_url": "https://mackerel.io/orgs/Hatena-org/alerts/5ryhYHNjTSQ",
   "monitor_name": "hatena.blog/404"
 }
}

Expression monitoring will no longer generate UNKNOWN alerts due to problems with Mackerel

Previously, there were cases of UNKNOWN alerts being triggered in expression monitoring due to problems with Mackerel, but there was no way for users to deal with this, so unnecessary alerts were being triggered.

We have therefore made adjustments so that UNKNOWN errors in expression monitoring will not be triggered if caused by problems with Mackerel in the future.

Please note that alerts not caused by problems with Mackerel, such as when the metric being monitored no longer exists, will continue to be triggered.

Further, please be aware that, due to this adjustment, if you refer to an alert that previously recorded the message “Either there is an error with the monitoring target''s expression or a time-out has occurred during value acquisition.” it will actually be replaced with the message “There is an error with the monitoring target's expression.” even if the UNKNOWN alert was caused by problems with Mackerel.

Units will now be displayed for custom metrics on role graphs

Units were not previously shown for custom metrics on role graphs, but now, units will be shown on “full-screen page role graphs” and “custom dashboard role graphs.” This will make it easier to check the units of metrics when looking at graphs.

APM and Traces have been added to functional components displayed at status.mackerel.io

When a fault occurs at Mackerel, status information is provided at status.mackerel.io. APM and Traces have been added to the functional components showing for which functions faults have occurred.

Component name Scope of effect
APM Viewing APM on the web console
Traces Posting and retrieving OpenTelemetry-compatible traces, viewing trace pages outside APM screen on web console, trace issues function

For information on other functional components, please see the FAQ below.

Where can I check information about faults? – Mackerel Support

terraform-provider-mackerel now supports specifying HTTP status codes that are allowed for external monitoring

You can select HTTP status codes deemed acceptable for external monitoring, but it is now also possible to specify codes by setting the value of expected_status_code when configuring monitoring with terraform-provider-mackerel. This is available via terraform-provider-mackerel v0.5.0.

This update was contributed by Mackerel user @fujiwara. Thank you!