Catching moment of build failure, "java -jar slave.jar" isn't running on slave, i.e. JVM with slave control crashed. Trying to restart slave soon after crash fails with "Could not allocate enough heap space". That's due to the fact that gcc/clang build still continuing, and apparently putting resources under good stress (I see loadavg of 100). That's how I got an idea that slave.jar may fail due to out of memory condition (note that I didn't see explicit Java OOM exception).
Catching moment of build failure, "java -jar slave.jar" isn't running on slave, i.e. JVM with slave control crashed. Trying to restart slave soon after crash fails with "Could not allocate enough heap space". That's due to the fact that gcc/clang build still continuing, and apparently putting resources under good stress (I see loadavg of 100). That's how I got an idea that slave.jar may fail due to out of memory condition (note that I didn't see explicit Java OOM exception).