Batch step running on a node other than the master fails

Running a batch command on another node takes several minutes and then fails. Attached example (all Windows, the echo command won't print):

stage('1') {
  node('uitest') {
    bat 'echo something'
  }
}

After 10 minutes the console output prompts this and the build fails:

ERROR: script apparently exited with code 0 but asynchronous notification was lost

In addition the system log gets these exceptions:

• java.lang.NoClassDefFoundError: Could not initialize class hudson.slaves.SlaveComputer
• hudson.remoting.Channel$CallSiteStackTrace: Remote call to JNLP4-connect connection from