Exception in thread "main" java.lang.AssertionError

[chsigg]

Description of the problem / feature request:

TensorFlow build recently started to hit the following error:

ERROR: /tmpfs/bazel_output/_bazel_kbuilder/f2d52ca1f092ccbe254cc98c3dc90790/external/bazel_tools/tools/jdk/BUILD:350:14: Action external/bazel_tools/tools/jdk/platformclasspath.jar failed (Exit 1): java failed: error executing command external/local_jdk/bin/java -XX:+IgnoreUnrecognizedVMOptions '--add-exports=jdk.compiler/com.sun.tools.javac.platform=ALL-UNNAMED' -cp ... (remaining 4 argument(s) skipped)
Exception in thread "main" java.lang.AssertionError: 
Could not find java.lang.Object on bootclasspath; something has gone terribly wrong.
Please file a bug: https://github.com/bazelbuild/bazel/issues
	at DumpPlatformClassPath.writeEntries(DumpPlatformClassPath.java:136)
	at DumpPlatformClassPath.writeClassPathJars(DumpPlatformClassPath.java:174)
	at DumpPlatformClassPath.dumpJDK8BootClassPath(DumpPlatformClassPath.java:77)
	at DumpPlatformClassPath.main(DumpPlatformClassPath.java:65)

Bugs: what's the simplest, easiest way to reproduce this bug? Please provide a minimal example if possible.

Sorry, I don't know of a simple repro. This is happening in one of the TensorFlow CI builds. Full logs are here

What operating system are you running Bazel on?

Ubuntu 16.04

What's the output of bazel info release?

$ bazel info release
release 3.7.2

$ java -version
java version "1.8.0_161"
Java(TM) SE Runtime Environment (build 1.8.0_161-b12)
Java HotSpot(TM) 64-Bit Server VM (build 25.161-b12, mixed mode)

The issue does not reproduce consistently (about 50% flaky) and started appearing when we switched the OS image to one with a newer CUDA version (I don't know exactly what else could have changed, I checked that the java version is the same).

[meteorcloudy]

@comius Any idea on this issue?

[hyeygit]

Is anyone looking at the issue? Could someone help find an assignee possibly?

[meteorcloudy]

Ping @comius @cushon

[meteorcloudy]

It's mostly like something is wrong with the JDK installed. I reply to the internal issue and hopefully we can find a way to debug this.

[sventiffe]

This issue stalled -- is it still a P1 (if so, please update the issue)? Else, if nobody is working on it, please bump it down to P2.

[comius]

The issue has been resolved internally by using JDK11.

There are inherently problems using JDK8, because tensorflow is using distinct_host_configuration=false configuration flag, which "melds" together JDK11 that's part of Bazel and JDK8 on the local system. Without flag compilation takes substantially longer time.