Cool beans.

If you're interested, myself (as well as others, I'm sure) would love to work on this.

Andrew Hulbert wrote:
With the single-node/docker I had to add a "sleep 3" after "hdfs
dfsadmin -safemode" returns too though before the "accumulo init"

I think having an Apache-owned docker for accumulo would rock! We have
made a few but having a canonical one would be great.

On 03/22/2016 02:47 PM, Michael Wall wrote:
Sven, did you ever put up a Dockerfile?  I've done something like this
waiting for HDFS to come out of safe mode.

echo "Starting Hadoop"
${HADOOP_HOME}/sbin/start-dfs.sh
echo "Waiting for hadoop to come out of safeMode"
local x=0
while [ $x -lt 1 ] do # try 5 times, for slow computers
  ${HADOOP_HOME}/bin/hdfs dfsadmin -safemode && x=5
  x=$(( $x + 1 ))
done

On Tue, Mar 22, 2016 at 2:41 PM, Josh Elser <josh.el...@gmail.com
<mailto:josh.el...@gmail.com>> wrote:

    Every time I use Docker, I seem to run into odd situations like
    this. Very frustrating.

    I would love to know what's happening and get this working
    reliably. Having an "Apache-owned" docker image for Accumulo would
    be nice (and would help us internally as well as externally, IMO).


    Andrew Hulbert wrote:

        +1 for Josh's suggestion.

        Not sure if its the same problem, but I too had to add some
        sleeps for
        single node accumulo startup scripts directly before and after
        init'ing
        accumulo. I dug in once and it seemed that the datanode needed
        more time
        to let other processes know the files in HDFS existed at
        /accumulo.
        Perhaps it was awaiting the exit from safe mode? Three seconds
        did it
        for me:
        
https://github.com/jahhulbert-ccri/cloud-local/blob/master/bin/cloud-local.sh#L109



        On 03/14/2016 01:49 PM, Josh Elser wrote:

            Well, I'm not sure wtf is going on, but I added some
            sleeps to your
            init.sh script (30seconds before starting ZooKeeper and before
            init'ing Accumulo) and now things worked just fine.

            I'm not seeing any Accumulo problems here, but rather the
            same Docker
            absurdities I run into every time I try to use it :). I
            also just
            upgrade to Docker 1.10.3 if that helps.

            Josh Elser wrote:

                So, I got it running, and ran into the same problem
                you described.

                The odd thing is that both the Accumulo Master and
                TabletServer seemed
                to be... stuck. They both claimed to be running, but
                they were both hung
                (unable to communicate). This ultimately caused the
                TabletServer to lose
                its ZooKeeper lock (and why the Shell didn't think it
                was running).

                I'm running into all sorts of docker issues ATM
                (trying to upgrade now,
                hoping it will help), but I'd recommend trying to
                increase the amount of
                heap for both the master and tserver from 128M to 256M in
                /accumulo-1.7.1/conf/accumulo-env.sh (the -Xmx value).

                Josh Elser wrote:

                    Ok! Thanks, I'll look for that init.sh script.

                    We're not too... docker-savvy here yet :), but we
                    can definitely help
                    out if we can find an automation script. Let me
                    get back to you.

                    Sven Hodapp wrote:

                        Hi Josh,

                        thanks for your answer.
                        Currently I haven't uploaded the Dockerfile...
                        If you want I'll upload
                        it!
                        (But currently it only is an debian with ssh,
                        rsync, jdk7, the
                        unzipped distributions of hadoop, zookeeper
                        and accumulo, and the set
                        of environment variables)

                        Currently, and for testing, I start all things
                        manually.
                        The steps I do is documented in /init.sh
                        within the docker container.
                        I've enshured that dfs and zookeeper are up
                        before starting accumulo.

                        Regards,
                        Sven

                        ----- Ursprüngliche Mail -----

                            Von: "Josh
                            
Elser"<<mailto:josh.el...@gmail.com>josh.el...@gmail.com>
                            An: "user"<user@accumulo.apache.org
                            <mailto:user@accumulo.apache.org>>
                            Gesendet: Montag, 14. März 2016 15:41:16
                            Betreff: Re: Accumulo 1.7.1 on Docker


                            Hi Sven,

                            I can't seem to find the source for your
                            Docker image (nor am smart
                            enough to figure out how to extract it).
                            Any chance you can point
                            us to
                            that?

                            That said, I did run your image, and it
                            looks like you didn't
                            actually
                            start Hadoop, ZooKeeper, or Accumulo.

                            Our documentation treats Hadoop and
                            ZooKeeper primarily as
                            prerequisites, so check their docs for
                            good instructions.

                            For Accumulo -

                            Configuration:
                            
http://accumulo.apache.org/1.7/accumulo_user_manual.html#_installation

                            Initialization:
                            
http://accumulo.apache.org/1.7/accumulo_user_manual.html#_initialization


                            Starting:
                            
http://accumulo.apache.org/1.7/accumulo_user_manual.html#_starting_accumulo




                            Sven Hodapp wrote:

                                Dear reader,

                                I'd like to create a docker image with
                                accumulo 1.7.1.
                                I've install it from scratch with
                                hadoop-2.7.2, zookeeper-3.4.8 and
                                accumulo-1.7.1.
                                I'm going though the installation and
                                every time I'll end up like
                                that:

                                root@deebd8e29683:/# accumulo shell -u
                                root
                                Password: ******
                                2016-03-13 13:18:31,627
                                [trace.DistributedTrace] INFO :
                                SpanReceiver
                                org.apache.accumulo.tracer.ZooTraceClient
                                was loaded successfully.
                                2016-03-13 13:20:31,955
                                [impl.ServerClient] WARN : There are no
                                tablet servers:
                                check that zookeeper and accumulo are
                                running.

                                Anybody got an idea whats wrong? I
                                have no idea anymore...

                                I'll currently share this docker image
                                on docker hub.
                                If you want to try it you can simply
                                start:

                                docker run -it themerius/accumulo
                                /bin/bash

                                All binaries and configs are in there.
                                In ./init.sh are my setup
                                steps.

                                Regards,
                                Sven




Reply via email to