November 14, 2018
JVM Suspension Time Monitoring
Until now, Instana only showed you the Garbage Collection time for JVMs. However, modern JVMs ship garbage collectors that are mostly concurrent, so that time does no longer “stop-the-world”. A much better metric is “suspension time”, which indicates how much application execution was delayed or interrupted. Instana uses a similar logic as https://github.com/giltene/jHiccup to measure this suspension time.
JVM Logging MDC support
Instana now populates the MDC of log4j 1&2 as well as logback. To add the Instana trace ID to your logs, just include the MDC variable
instana.trace.id in you log appender. Please refer to the documentation of your logger for details how to set this up.
Metadata Support for Oracle Public Cloud servers
Agents running on OPC will now report the metadata available in the Oracle metadata server, including the failure domain which will be reported as zone.
Instana now allows to define time windows to mute alerts of specific entities. This is for example useful when there is planned maintenance or load testing in parts of the monitored system.
New Python Instrumentation
- Instana now supports Java 11, as well as experimental support for Java 12 already.
- Improved Java Tracing for JAX WS calls, added support for tracing Netscape LDAP
- The agent will now also discover Java processes under windows which are not named “.exe”
- Lowered resource usage of the Kubernetes monitoring
- Consul sensor will no longer report false changes due to internal sorting of members
- JBoss Sensor will now wait for the server startup, so that it no longer produces “service not deployed” log messages.
- NGinx Sensor will now be able to parse configurations containing Lua rewrites
- Analyze View: Support filtering using “does not equal”, “does not contain” and “is not present” operators
- Service and Endpoint Mapping: improved tag extraction and naming for several components / span types
- Python Sensor: Add support for Path Templates.
- Integrations: Test call button to trigger a test message.
- Cassandra DSE 3.11 can now be monitored correctly
- CXF calls using Async Http Client now show up correctly
- MS-SQL Sensor: Fixed issue preventing metric collection on certain installations
- Fixed tiny memory leak that grew over time when a lot of services started and stopped.
- Fixed Java ForkJoinPool Tracing in Java 10+
- Fixes edge case where a JVM was incorrectly identified as Elasticsearch
- Analyze View: Fixed trace and call count
- Analyze View: Fixed showing traces and calls for the “Unrecognized” service
- External services called via HTTP no longer have an incorrectly detected technology associated to them
- We are removing custom rule/issue support on website entities for environments that currently have no active configurations, in preparation for the improved alerting mechanism that will be available once the end-user monitoring rework is delivered. If your team has unique end-user monitoring alerting needs, please reach out to us.