Selenium grid v3.10.0 doesn't start at docker container

[VitarLaeda]

Meta -

OS:
Docker container
Selenium Version:
3.10.0

Grid

Expected Behavior -

Correct start of selenium grid in docker container
With selenium 3.9.1 all works correctly

Actual Behavior -

Container doesn't start

Exception in thread "main" java.lang.NullPointerException
at org.openqa.grid.internal.cli.GridHubCliOptions.lambda$defaults$0(GridHubCliOptions.java:52)
at com.beust.jcommander.JCommander.initializeDefaultValue(JCommander.java:662)
at com.beust.jcommander.JCommander.initializeDefaultValues(JCommander.java:356)
at com.beust.jcommander.JCommander.parse(JCommander.java:339)
at com.beust.jcommander.JCommander.parse(JCommander.java:319)
at org.openqa.grid.internal.cli.GridHubCliOptions.parse(GridHubCliOptions.java:42)

Steps to reproduce -

link to GitHub repository or

config :

#!/bin/bash

echo "
{
  \"host\": null,
  \"port\": 4444,
  \"prioritizer\": null,
  \"capabilityMatcher\": \"org.openqa.grid.internal.utils.DefaultCapabilityMatcher\",
  \"throwOnCapabilityNotPresent\": true,
  \"newSessionWaitTimeout\": $GRID_NEW_SESSION_WAIT_TIMEOUT,
  \"jettyMaxThreads\": $GRID_JETTY_MAX_THREADS,
  \"nodePolling\": $GRID_NODE_POLLING,
  \"cleanUpCycle\": $GRID_CLEAN_UP_CYCLE,
  \"timeout\": $GRID_TIMEOUT,
  \"browserTimeout\": $GRID_BROWSER_TIMEOUT,
  \"maxSession\": $GRID_MAX_SESSION,
  \"unregisterIfStillDownAfter\": $GRID_UNREGISTER_IF_STILL_DOWN_AFTER
}"

Dockerfile :

FROM netflixoss/java:8

#------------------------------------------------
# Customize sources for apt-get
#------------------------------------------------

RUN  mkdir -p /opt/selenium \
    && wget --no-verbose http://selenium-release.storage.googleapis.com/3.10/selenium-server-standalone-3.10.0.jar \
    -O /opt/selenium/selenium-server-standalone.jar

RUN useradd seluser --shell /bin/bash --create-home \
  && usermod -a -G sudo seluser \
  && echo 'ALL ALL = (ALL) NOPASSWD: ALL' >> /etc/sudoers \
  && echo 'seluser:secret' | chpasswd

#------------------------
# Selenium Configuration
#------------------------

EXPOSE 4444
EXPOSE 5555

ENV GRID_NEW_SESSION_WAIT_TIMEOUT 0
ENV GRID_JETTY_MAX_THREADS -1
ENV GRID_NODE_POLLING  5000
ENV GRID_CLEAN_UP_CYCLE 5000
ENV GRID_TIMEOUT 30000
ENV GRID_BROWSER_TIMEOUT 60000
ENV GRID_MAX_SESSION 5
ENV GRID_UNREGISTER_IF_STILL_DOWN_AFTER 30000

COPY generate_config /opt/selenium/generate_config
COPY entry_point.sh /opt/bin/entry_point.sh
RUN chmod a+rx /opt/bin/entry_point.sh
RUN chmod a+rx /opt/selenium/generate_config

RUN chown -R seluser /opt/selenium

USER seluser

CMD ["/opt/bin/entry_point.sh"]

entry_point.sh :

#!/bin/bash

ROOT=/opt/selenium
CONF=$ROOT/config.json

. $ROOT/generate_config >$CONF
echo "starting selenium hub with configuration:"
cat $CONF

if [ ! -z "$SE_OPTS" ]; then
  echo "appending selenium options: ${SE_OPTS}"
fi

function shutdown {
    echo "shutting down hub.."
    kill -s SIGTERM $NODE_PID
    wait $NODE_PID
    echo "shutdown complete"
}

java ${JAVA_OPTS} -jar /opt/selenium/selenium-server-standalone.jar \
  -role hub \
  -hubConfig $CONF \
  ${SE_OPTS} &
NODE_PID=$!

trap shutdown SIGTERM SIGINT
wait $NODE_PID

[akeshmiri]

+1, same configuration, same error

[akeshmiri]

I think I found the root cause of issue after code investigation.
The problem is that in generate_config.sh inside docker there is a parameter "prioritizer": null,

In selenium, GridHubCliOptions -> defaults method, line 52 we have

  return map.containsKey(option) ? map.get(option).toString() : null;

Short term solution is to remove "prioritizer": null,, for long term, I think selenium code should be updated to handle null values.

[barancev]

The fix will be shipped as part of release 3.11