在 Jenkins 上构建 Android 项目时，Gradle build 守护进程意外消失(它可能已经被杀死或者已经崩溃)

小开

EDIT Looks like there has been a few changes with the new versions of Gradle.

Since 3.0 you should not disable the daemon on your CI anymore

[We] recommend using [the daemon] for both developers' machines and Continuous Integration servers.

However, if you suspect that Daemon makes your CI builds unstable, you can disable it to use a fresh runtime for each build since the runtime is completely isolated from any previous builds.

PREVIOUS ANSWER

It's recommended to turn off daemon on any CI server. use this option to disable it

--no-daemon

小开

Gradle build daemon disappeared unexpectedly in many cases mean gradle itself or even java crashed. In my case it was java. Fill bugreport: https://bugzilla.redhat.com/show_bug.cgi?id=1408857

Look at files named like: hs_err_pid%p.log where %p is PID of process in directory from what you are run gradle task.

Update: It looks like gradle itself issue. In my case because of use native jansi. In issue provided workaround:

ln -sb /dev/null /home/pasha/.gradle/native/jansi/1.17.1/linux64/libjansi.so

小开

I have tried the --no-daemon solution but my build continued to fail with the same DaemonDisappearedException.

I solved this by increasing the RAM of the server I am running Jenkins upon. In AWS EC2 that meant having to increase the EC2 instance type which results to an increase RAM.

小开

It seems like it is a memory related issue. Nevertheless, disabling the daemon as suggested by Oleg does seem to help.

Use

org.gradle.daemon=false

in

gradle.properties

either in ~/.gradle folder or the project's folder.

Ref: https://docs.gradle.org/current/userguide/gradle_daemon.html#sec:disabling_the_daemon

小开

After getting this crash I tried several things to get the GradleDaemon to stop running on my CI server. None of which worked.

I found an answer on a gradle.org forum which suggested that the GradleDaemon would always run anyway. The --no-daemon flag would just make it run for this specific build rather than staying on indefinitely.

If you specify JVM arguments that require forking, Gradle will fork a new JVM. Regardless of whether or not you want a daemon process, the class that runs is called GradleDaemon. The --no-daemon switch should cause the forked process to be single use instead of a long running daemon process, but it's still going to run the GradleDaemon class.

Source: https://discuss.gradle.org/t/no-daemon-switch-ineffective-if-jvm-settings-cause-new-fork/14919/5

I may be reading this wrong and I can't vouch for the validity of the answer, but I think the cause of this error is just a lack of memory for Gradle. As it is always going to run the GradleDaemon.

So I added

org.gradle.jvmargs=-Xmx1024m

to my ~/.gradle/gradle.properties file and it no longer gave me that error.

小开

I was using Android Studio in Windows 7 and then this error appeared. What worked for me is killing Java.exe from Windows TaskManager.

小开

In our case the issue was caused by the CI server passing environment variables with non-ascii characters (i.e. in the names of commit authors).

Adding file.encoding=utf-8 to Gradle properties fixed the issue immediately.

小开

gradle -Dorg.gradle.jvmargs=-Xmx1536m assembleDebug

Or add org.gradle.jvmargs=-Xmx1536m to gradle.properties file.

小开

Sometimes just doing Build -> Clean Project works - Give that a try before you start with the other gradle file changes.

小开

No one else may run into this, as it's quite silly, but...

My issue was a strange character being present in my commit message...I had copied a previous commit message from gitlab which contained an emoji and pasted that into the title of a merge request, instead of the normal :bug: syntax.

akru's answer helped point me in the right direction 🙏

小开

In my case, I was upgrading my android studio project and I use ZelixKlassMaster for obfuscating my code the issue wa that, I had my class-path on Zelix set to 27 but my project was using android 28 Hope this will help any future comers the way I debugged this was my seed file was printing out this error

ERROR: Invalid classpath in "classpath" statement at line 69 : "C:\Users\Rab\AppData\Local\Android\Sdk\platforms\android-27\android.jar" is not a valid path.

小开

Wow, in my case closing the Android Studio and re-opening it worked just fine and the error was gone. :)

小开

Go to /gradle.properties then remove org.gradle.jvmargs=-Xmx1024m and if org.gradle.jvmargs=-Xmx1024m not available then Add this code in your /gradle.properties.

小开

I had the same issue, in the end there was a missing argument in the gradle file

Basically, this was a kotlin project, in which some experimental coroutine features were used. The class which used it was marked with

@OptIn(FlowPreview::class)

This requires adding the following argument in the build.gradle, although it was running just fine in android studio locally

kotlinOptions {
jvmTarget = "1.8"
freeCompilerArgs = [ "-Xopt-in=kotlin.RequiresOptIn"]
}

Spent a lot of time looking for this one

小开

Most of the time you only need to restart Android Studio and it should work. Also you can do the following: File -> Sync Project with Gradle Files and then File -> Invalides Cachse / Restart.

小开

Before assemble task.

./gradlew --status

check daemon status.

Then

./gradlew --stop

to stop the daemon.

Use

./gradlew assemblerelease  --no-daemon -Dkotlin.compiler.execution.strategy="in-process"

to disable daemon.

小开

Add in gradle.properties file.

org.gradle.daemon=true <br>
org.gradle.jvmargs=-Xmx1024m <br>
android.useDeprecatedNdk=true <br>
android.useAndroidX=true <br>
android.enableJetifier=true <br>
file.encoding=utf-8

小开

i had the same issue added these in the gradle.properties file

android.useAndroidX=true
android.enableJetifier=true
org.gradle.jvmargs = -Xmx2g

小开

I had the same problem and after a long fight, I deleted some files and got some free space in memory. then restarted android studio, now it works perfectly.

小开

I was experiencing the same issue in GitHub Actions. As other answers mention, it seemed to be a memory issue.

The following settings worked nicely for me with the GitHub Ubuntu runner (no need to disable the deamon):

# ...
env:
GRADLE_OPTS: -Dkotlin.incremental=false -Dorg.gradle.jvmargs="-Xmx4g -XX:MaxMetaspaceSize=2g -XX:+HeapDumpOnOutOfMemoryError -Dfile.encoding=UTF-8"
jobs:
build:
runs-on: ubuntu-latest
steps:
# ...
- name: Build app
run: ./gradlew assembleRelease --stacktrace

小开

In my case, removing org.gradle.parallel=true did the trick.

My gradle.properties for your reference:

org.gradle.daemon=false
org.gradle.jvmargs=-Xmx3g -XX:MaxMetaspaceSize=768m -XX:+HeapDumpOnOutOfMemoryError -Dfile.encoding=UTF-8
kotlin.compiler.execution.strategy=in-process

PS: I'm using this configuration for running Android Workflows with GitHub Actions (ubuntu-20.04)

小开

In my case, I was using Docker for running the Gradle tasks from Github actions locally.

I had to increase docker memory, and it worked!

小开

I got the same error in Docker, and I tried to set org.gradle.daemon=false and all the above ways, but it didn't work . At last, I update the gradle to version 6.5 and the android gradle plugin to version 4.1.0. then the error disappeared.

小开

It appears this type of problem (Gradle build daemon disappeared unexpectedly (it may have been killed or may have crashed)) is becoming more common as people move to GitHub workflows and GitHub Actions.

We never saw this problem when building locally (command line or Android Studio), or when building on our Jenkins build server. But as soon as we started testing builds through GitHub Workflows/Actions this type of build error would happen intermittently.

Every project is different, so it appears there are several solutions that may work. After a lot of experimenting with our builds, the only parameter that reliably fixed this problem was '-XX:MaxMetaspaceSize'.

GRADLE_OPTS: -Dorg.gradle.jvmargs="-XX:MaxMetaspaceSize=1g"

No changes to build.gradle, gradle.properties, gradle-wrapper.properties, etc. were needed in our case. Just the single GRADLE_OPTS line in our GitHub workflow yaml. This explanation from the Java docs was the most helpful (avoiding unbounded growth of class metadata space).

In previous releases of Java Hotspot VM, the class metadata was allocated in the so-called permanent generation. Starting with JDK 8, the permanent generation was removed and the class metadata is allocated in native memory. The amount of native memory that can be used for class metadata is by default unlimited. Use the option -XX:MaxMetaspaceSize to put an upper limit on the amount of native memory used for class metadata.

Source: https://docs.oracle.com/en/java/javase/17/gctuning/other-considerations.html

Working example of GitHub Workflow:

jobs:
build:


runs-on: ubuntu-latest


env:
GRADLE_OPTS: -Dorg.gradle.jvmargs="-XX:MaxMetaspaceSize=1g"


steps:
- uses: actions/checkout@v2
with:
ref: 'master'
fetch-depth: 0


- name: Set up JDK 11
uses: actions/setup-java@v2
with:
java-version: '11'
distribution: 'temurin'
cache: gradle


- name: Gradle Build
uses: gradle/gradle-build-action@v2
with:
arguments: build -x lint appDistributionUploadRelease

小开

This is how it worked for me:

Enable the gradle deamon (this is the default with gradle 7).

org.gradle.daemon=false

Install Jenkins Plugin: https://plugins.jenkins.io/gradle-daemon/

The Plugin will prevent to kill the deamon if another build is still using it.