Announcements

#mackerelio

Lambda and ECS Cluster hosts can now be auto retirement, and more

Hello! This is Igarashi (id:masarasi) from the Mackerel CRE team. Here are the details on our latest update.

Lambda and ECS Cluster hosts can now be auto retirement

AWS integration now allows Lambda and ECS Cluster hosts to be auto retirement. Please make use of this feature.

Auto retirement settings

Improved the display of service names when filtering host list by cloud integration

When filtering by cloud integration on the host list screen, multiple different services were being displayed under the same name. Therefore, the following changes have been made to the display of service names.

Cloud platform Before change After change
AWS Batch AWS Batch
Azure Database for MySQL Database for MySQL Flexible Server
Database for MySQL Single Server
Database for PostgreSQL Database for PostgreSQL Flexible Server
Database for PostgreSQL Single Server

Filter by cloud integration

Email addresses for inviting members are no longer case-sensitive

Previously, when inviting members to join an organization, the case of the email address entered had to match the email address registered to the account for the invitee to be able to join the organization from the link in the invitation email. This update makes it possible for invitees to join an organization even if the case of the email addresses do not match.

Search results in the monitoring rule list are now retained after the monitoring rules details screen is opened

Previously, when searching for a monitoring rule using the search box at the top of the monitoring rule list screen, opening the monitoring rule details screen from the search results, and returning to the monitoring rule list screen, the search results would be reset. The original search results will now be retained.

Fixed a bug in mackerel-plugin-squid

Fixed a bug in mackerel-agent-plugins v0.54.0 and later that prevented mackerel-plugin-squid graphs from displaying properly.

Latest version: mackerel-agent-plugins v0.88.3

Please note that additional steps may be required when updating to the latest version, depending on your usage of mackerel-agent-plugins.

Usage Additional steps
If you are using v0.54.0 or earlier Not required
If you used v0.54.0 or earlier, but have updated to v0.54.0 or later Required
If you are using v0.54.0 or later from the start Required

If you need to perform additional steps, please follow the below steps after updating.

  1. Comment out the mackerel-plugin-squid setting in mackerel-agent.conf and restart mackerel-agent. If mackerel-plugin-squid is configured on multiple hosts, do this for all hosts.
  2. On the Graph definition screen, expand graph definitions containing the name squid and click "delete this graph" to delete the graph definition. Repeat this step until there are no more graph definitions containing the name squid. (User-submitted custom metrics containing the name squid do not need to be deleted.)
  3. Restore the mackerel-plugin-squid setting in mackerel-agent.conf and restart mackerel-agent.
  4. Check if the following graph definitions are displayed on the Graph definition screen.

If functioning normally, mackerel-plugin-squid will create the following 6 graph definitions.

Graph definition Graph name Metric name
custom.squid.cpu_usage_ratio.5min.* Squid CPU Usage Ratio (5min) CPU Usage Ratio
custom.squid.cache_storage_usage.* Squid Cache Storage Usage Memory capacity (used)
Swap capacity (used)
custom.squid.cache_hit_ratio.5min.* Squid Client Cache Hit Ratio (5min) Byte Ratio
Request Ratio
custom.squid.requests.* Squid Client Requests Requests
custom.squid.file_descriptor_usage.* Squid File descriptor usage Files queued for open
Store Disk files open
Reserved number of file descriptors
Available number of file descriptors
Number of file desc currently in use
Largest file desc currently in use
Maximum number of file descriptors
custom.squid.memory_account_for.* Squid Memory accounted for memPoolFree calls
memPoolAlloc calls
  • If you are using mackerel-agent-plugins v0.54.0 or earlier, only custom.squid.cache_hit_ratio.5min.* and custom.squid.requests.* will be available. Other metrics will be posted after updating to the latest version.
  • If you are using v0.54.0~v0.88.2, the above graph definitions will be created with different graph names or with %1 in the metric name. The graphs will still be displayed, but an error log will be output to mackerel-agent, or the graphs will not be displayed as intended by the plugin.

Improved APM function

The number of spans is now displayed on the trace details screen

The total number of spans in a trace is now displayed on the trace details screen that appears when a trace in the trace list is selected. Furthermore, the value of service.namespace is displayed in a tooltip when mousing over a service name.

Trace details screen

Improved trace details screen UI

  • Spans now change color when selected.
  • Partial scrolling is no longer required in the span attribute display area.

Selected span

Improved trace filter conditions input completion function behavior

When using Attribute or Resource Input Completion in the filter conditions of a trace, the suggestions were not always displayed correctly. This has been fixed. Furthermore, suggestions are now displayed even if no characters are entered.

Changed aggregation method on HTTP server and database tabs

The aggregation of "Root" in the HTTP server tab and "Query" in the database tab of the APM screen now excludes content exceeding 2,048 characters. Queries with the same content in the first 2,048 characters will be aggregated as the same query, even if their contents differ after the first 2,048 characters, so please take note of this as a specification change. The purpose of this change is to improve aggregation processing performance.

check-systemd-journal, which can monitor systemd journal logs, is now available on Mackerel Labs

In recent years, systemd journal logs have become common in Linux systems, but as the check-log plugin does not allow direct monitoring of systemd journal logs, it is necessary to export logs as separate log files using rsyslog, etc. check-systemd-journal is a plugin that enables direct monitoring of journal logs. If you wish to use this plugin, download and install the package for your OS from the release page as it is not included in the official plugin collection. For information on how to use check-systemd-journal, please refer to the README.

Software made available on Mackerel Labs is experimental and is not supported. Please note that we will in principle be unable to respond to inquiries regarding such software.

Changed terraform-provider-mackerel SDK to Terraform plugin framework

The new implementation of the terraform-provider-mackerel SDK based on the Terraform plugin framework is now the default, and the old implementation is available on an opt-in basis. You can continue to use the legacy implementation by setting the MACKEREL_LEGACY_SDK environment variable to 1 or true.

Latest version: terraform-provider-mackerel v0.6.0

mackerel-agent-plugins currently includes several AWS-related metric plugins, but we plan to remove these plugins in the near future. This is because these plugins can generally be replaced with AWS integration, and the Mackerel team is emphasizing on the monitoring experience using AWS integration.

The plugins that will no longer be included are listed below. mackerel-plugin-aws-ec2-ebs will continue to be included as it cannot be easily replaced with AWS integration.

Please note that even after these plugins are removed, you will still be able to download and use them from their respective repositories. They will however no longer be supported.

[Future update] Time will no longer be displayed when mousing over a metric with a granularity exceeding 1 day

The granularity of metrics in Mackerel graphs changes depending on the period displayed, but the granularity of the metrics displayed in the graphs becomes more than 1 day apart when the period reaches the 3 month mark. When mousing over a graph, a time of 09:00 is currently displayed in addition to the date, but the actual value is not the value as of 09:00. This is confusing, and consequently, this time will no longer be displayed.

Metric with a granularity of 1 day

[Future update] AWS WAF Classic will no longer be supported in AWS integration

Currently, AWS integration can be used to monitor AWS WAF Classic and AWS WAF, but around September 2024, AWS announced the discontinuation of AWS WAF Classic on September 30, 2025.

In order to respond to the discontinuation of AWS WAF Classic, Mackerel will also discontinue support for AWS WAF Classic in AWS integration. The detailed schedule will be announced once it has been finalized, but please note that AWS WAF Classic support may be discontinued before September 30, 2025, the date on which AWS WAF Classic will be discontinued.

Even after AWS WAF Classic is no longer supported, you can continue to monitor AWS WAF. There may be some users who have already migrated to AWS WAF, but users who area still monitoring AWS WAF Classic should migrate to AWS WAF.

Thank you for your understanding.

[Event] Various events for learning about observability in detail will be held!

July 30 (Wed) 12:00~@ online Hatena Engineer Seminar #34 Achieving and Applying Observability

The theme of this event is "observability," which has become indispensable in the operation of increasingly complex IT systems. Engineers from the Mackerel development team and SREs from the system platform team will share practical knowledge on how to achieve observability in order to support service operations and rapid problem resolution.

hatena.connpass.com

7 August (Thu) 19:00~@ Hatena Tokyo Office Introduction to Observability with Japanese Services [online stream available]

Based on the book "Introduction to Observability with Japanese Services," this event will introduce practical approaches that developers can implement and utilize on their own. This event will delve into the specifics of how to face a field that is often considered to be "something to be dealt with by people who operate services" from the perspective of people who write code on a daily basis.

mackerelio.connpass.com