Build 147

2019-02-26

On-Premises Release Notes

For an upgrade to the new version please follow the Packaged Based Operations manual.

Deprecation Notice

The new EUM functionality relies on a completely new data format, so after the update you will not have historic EUM data. To access your old EUM data, you can still access the old dashboards at the following path in this release: /#/websites. Be aware that with the next release the old EUM data will not be accessible anymore.

Cumulative Features from Release #146 and #147

New Website Monitoring

This release introduces a completely revamped website monitoring experience! Not only does it visually align website monitoring with application perspectives, it also supports…

  • full insights into all website data for every page,
  • detailed resource and HTTP insights,
  • geographic breakdowns,
  • advanced analytics capabilities for website monitoring data,
  • selectable and real (non-averaged) percentiles for all time metrics,
  • a dedicated view for page load data,
  • React error boundary integration,
  • optional identification of users,
  • better backend correlation experience,
  • better guidance when encountering problems and deprecations.

No need to change anything, just head to the Websites area within the product!

Website Overview Dashboard

Increase retention time for 1s-rollups to 24h

We have increased the retention time of 1s rollups from 10 minutest to 24 hours for all customers.

All metrics will be available this way, both via API and the Instana UI.

Toggle Built-in Health Rules

Instana’s built-in health rules, used to explain the health of monitored systems in real-time, are now exposed and can be selectively toggled on/off. Further documentation can be found here.

Micrometer

Instana now supports monitoring of Micrometer metrics. Documentation can be found here.

We simplified analyzing calls that relate to an application, service or endpoint issue. An “Analyze Calls” button was added to the details page of an issue which navigates to the Analyze dashboard with all useful filters already set.

Issue Analyze Calls Link

AWS DynamoDB health rules

Added two health rules for critical reads and writes consumed/provisioned ratio.

Other Features

  • Node.js: Record koa-router routes as path templates when tracing HTTP entries.
  • Node.js: Add HTTP query params for HTTP exits

Improvements

  • Support for Kubernetes 1.11, 1.12, 1.13
  • More efficient AWS metrics gathering. Please change the permission from cloudwatch:GetMetricStatistics to cloudwatch:GetMetricData.
  • Configurable polling intervals for AWS services. See individual AWS services for details: e.g. DynamoDB, ElastiCache or RDS.
  • Google Marketplace for Kubernetes support. Read more on that in our blog: Instana available on Google Kubernetes Engine.
  • Google App Engine (GAE) support
  • Azure SQL Database support
  • Azure API Management support
  • Azure CosmosDb Metrics are now displayed per collection and per operation based on data model.
  • The Kafka dashboard was improved and the latency metrics are now correct.
  • Application perspectives: Top traces can be shown by trace count, latency and error rate with links to Analyze View.
  • Trend detection on applications, services and endpoints: Reduced number of false positives by also incorporating the absolute change in value.
  • etcd sensor now can connect to secure endpoints using custom certificates.
  • Initial support for monitoring processes inside LXC containers.
  • Initial support for monitoring processes inside containerd containers.
  • Improved discovery for Oracle DB on Windows.
  • Reduced memory usage of Kubernetes Sensor.
  • Support for PHP 7.3.
  • SpringBoot sensor will now discover non-fat jar deployments.
  • Java HornetQ tracing is now supported for even more deployment models of HornetQ.
  • Nginx Sensor will now walk symlinked directories for config parsing.
  • PHP Tracing supports now muslc, for example on Alpine Linux.
  • Improved Dropwizard dashboard.

Fixes

  • Application perspectives: Fixed empty endpoint dashboard when endpoint name is too long.
  • Extended support for linking services, endpoints, and calls to clusters.
  • CLR Sensor will no longer stop working when encountering “mem_po” values.
  • Improved logging for several sensors.

Known Issues

  • Website monitoring caching statistics before 2019-01-09 are not correct.
  • Geographic country breakdowns are currently only enabled for Germany and the United States. State-level data can also be missing for specific IP ranges, e.g. company internal networks. An explanation for these cases will be added to the product soon.