GC overhead limit exceeded

[dovogt]

Gradle 5.0 changed the memory settings. The following error occurs during execution:

Execution failed for task ':spotbugsMain'.

A failure occurred while executing com.github.spotbugs.internal.spotbugs.SpotBugsRunner
GC overhead limit exceeded

I use it together with the ru.vyarus: gradle-quality-plugin

[KengoTODA]

Hi, do you mean this change? Could you share mcve to reproduce your problem?

[dovogt]

Sorry for my late reply. Enclosed I send you the log file with debug output.
Which memory switch is crucial, I could not find out.
[
gradle.log
](url)

[KengoTODA]

Thanks for your share! I also found nothing from this Gradle log 😭

1. Could you try to specify larger -Xmx option via JAVA_OPTS or GRADLE_OPTS environment variable?
2. Could you reproduce this issue with spotbugs-gradle-plugin version 1.6.4?

[dovogt]

Hy,

I have the latest version in use and started the task with --scan. The memory in the gradlew.bat I have increased to 1024MB

https://scans.gradle.com/s/vciemyka5mvk6

Maybe it helps us to continue to find the error :-(

@rem Add default JVM options here. So you can use JAVA_OPTS and GRADLE_OPTS to pass JVM options to this script.
set DEFAULT_JVM_OPTS = "- Xmx1024m"

[oehme]

You're changing the memory for the Gradle daemon, which won't do anything, because Spotbugs is run in a different VM. Change the memory settings of Spotbugs intead using spotbugs.jvmArgs.

[shevek]

Running spotbugs in a separate JVM is a really bad idea from the system resource point of view. It's fine for single threaded builds on a large system, but no good for parallel builds, where we end up spawning a LARGE NUMBER of worker JVMs, and the system runs out of memory at OS level, and the entire build dies.

Please can we make spotbugs in instance-safe, in-process, using the resources allocated to the primary Gradle JVM, so that we can have sane resource allocation and reliable builds.

[jscancella]

@dovogt please reopen if this is still an issue