Closed klamouri closed 1 month ago
Hi @klamouri We're aware of this issue, and currently the workaround is one of:
HttpUrlConnection
instrumentation module (see below). Note that doing this can result in lost/incomplete traces depending on the HTTP client being used.To disable the instrumentation module, add the following to your yml config file:
class_transformer:
com.newrelic.instrumentation.httpurlconnection:
enabled: false
We will be publishing a fix for this soon.
The fix will be in the 8.11.0 release. Here's the related issue https://github.com/newrelic/newrelic-java-agent/issues/1838
Java agent 8.11.0 is now out which fixes this issue: https://github.com/newrelic/newrelic-java-agent/releases/tag/v8.11.0
The 8.10.0 release notes/changelog have also been updated to call out the bug in that version.
Description
Our app thread pool started filling up after no apparent changes on our side. We traced it back to the new relic agent version 8.10.0
Indeed we used to pull the latest version when server boots and since March 30th we started seeing our applications running out thread from our thread pool. After few tests and thread dumps we realized that there seems to be a deadlock in the agent It seems it's only affecting HTTP clients using
sun.net.www.protocol.https.HttpsURLConnectionImpl
We had hundreds of threads hanging with this stacktrace.Expected Behavior
Threads should not hang
Your Environment
Java 17 (Corretto) Spring Boot 3.2.4 New Relic Agent 8.10.0
Additional context
After downgrading the agent version back to 8.9.0 no more thread hang and our http calls are way faster as visible in New Relic UI