You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@drill.apache.org by Harri Kinnunen <Ha...@hitsaamo.fi> on 2013/09/24 22:10:20 UTC
AWS AMI for Drill M1
All,
talked with Tim on the hangout about setting up a standalone instance of drill and publishing that as an Amazon AMI. The purpose of this is just to experiment with AWS so that Tim can carry forward with the AWS cluster setup. If I can later clear up the JAVA_HOME thing mentioned below, we could point to it as an alternative for a easy environment for trying Drill out.
So ...
The AMI is "ami-1343167a" and its name is "Drill M1 standalone", but you should find it using the ID in "EC2 / Images / AMIs". I launched it as t1.micro instance, and at least it was able to start up sqlline. Amazon says t1.micro has 615mb RAM.
Running it:
Make sure you're pointing to the correct Java version:
$ sudo alternatives --config java
(pick the one that points to Java 1.7)
Somehow the JAVA_HOME got messed up after the first-ever execution of sqlline. So you need to:
$ JAVA_HOME=/usr/bin/java
That's all there is to it.
$ cd /home/ec2-user/apache-drill-1.0.0-m1
$ bin/sqlline
Installation log … Sort of:
$ cd /home/ec2-user
$ sudo yum install java-1.7.0-openjdk-devel
(I'm not sure if JDK is needed or just the JRE, so I put the JDK. This can also be the source of the java problems above…)
$ curl http://people.apache.org/~jacques/apache-drill-1.0.0-m1.rc3/apache-drill-1.0.0-m1-binary-release.tar.gz > apache-drill-1.0.0-m1-binary-release.tar.gz
$ tar -zxf apache-drill-1.0.0-m1-binary-release.tar.gz
Cheers,
-Harri
Re: AWS AMI for Drill M1
Posted by Harri Kinnunen <Ha...@hitsaamo.fi>.
Hi,
still tweaking around with this one …
Tried with Oracle JRE. The previous version was on OpenJDK JDK.
Oracle actually seems to take into account the XX:MaxDirectMemorySize java parameter. And thus sqlline refuses to start on Amazon Micro Instance.
However, without the parameter sqlline starts fine. And at least simple queries on test data execute.
Is there a reason why we have the parameter there? The parameter talks about "max" direct memory. I would understand "min" (as a sort of system requirement). But "max"? If we don't have it, will it hog all resources, or something even more sinister?
I'd rather have this on Oracle (I guess it should be considered as the "reference" JRE…).
Cheers,
-Hari
On 25.9.2013, at 0.04, Timothy Chen <tn...@gmail.com> wrote:
> Ya I had to set the DRILL_LOG_DIR to somewhere in /usr/local/ in my Whirr
> scripts setup.
>
> Tim
>
>
> On Tue, Sep 24, 2013 at 2:02 PM, Ted Dunning <te...@gmail.com> wrote:
>
>>
>> On Tue, Sep 24, 2013 at 1:10 PM, Harri Kinnunen <
>> Harri.Kinnunen@hitsaamo.fi> wrote:
>>
>>> JAVA_HOME=/usr/bin/java
>>
>>
>>
>> Also has a problem the log directory not being writable.
>>
>>
Re: AWS AMI for Drill M1
Posted by Timothy Chen <tn...@gmail.com>.
Ya I had to set the DRILL_LOG_DIR to somewhere in /usr/local/ in my Whirr
scripts setup.
Tim
On Tue, Sep 24, 2013 at 2:02 PM, Ted Dunning <te...@gmail.com> wrote:
>
> On Tue, Sep 24, 2013 at 1:10 PM, Harri Kinnunen <
> Harri.Kinnunen@hitsaamo.fi> wrote:
>
>> JAVA_HOME=/usr/bin/java
>
>
>
> Also has a problem the log directory not being writable.
>
>
Re: AWS AMI for Drill M1
Posted by Ted Dunning <te...@gmail.com>.
On Tue, Sep 24, 2013 at 1:10 PM, Harri Kinnunen
<Ha...@hitsaamo.fi>wrote:
> JAVA_HOME=/usr/bin/java
Also has a problem the log directory not being writable.
Re: AWS AMI for Drill M1
Posted by Ted Dunning <te...@gmail.com>.
We would be happy to echo this blog on the MapR blog as well.
On Tue, Sep 24, 2013 at 1:20 PM, Michael Hausenblas <
michael.hausenblas@gmail.com> wrote:
>
> Awesome! One thing less on my to do list :)
>
> Harri, how about writing a short blog post about it? I'm maintaining
> http://drill-user.org/ and this would make a great post, there.
>
> Also, if anyone else here on the mailing list wants to use this blog to
> report on progress and/or experiences with Drill, please let me know!
>
> Cheers,
> Michael
>
> --
> Michael Hausenblas
> Ireland, Europe
> http://mhausenblas.info/
>
> On 24 Sep 2013, at 21:10, Harri Kinnunen <Ha...@hitsaamo.fi>
> wrote:
>
> > All,
> >
> > talked with Tim on the hangout about setting up a standalone instance of
> drill and publishing that as an Amazon AMI. The purpose of this is just to
> experiment with AWS so that Tim can carry forward with the AWS cluster
> setup. If I can later clear up the JAVA_HOME thing mentioned below, we
> could point to it as an alternative for a easy environment for trying Drill
> out.
> >
> > So ...
> >
> > The AMI is "ami-1343167a" and its name is "Drill M1 standalone", but you
> should find it using the ID in "EC2 / Images / AMIs". I launched it as
> t1.micro instance, and at least it was able to start up sqlline. Amazon
> says t1.micro has 615mb RAM.
> >
> >
> > Running it:
> > Make sure you're pointing to the correct Java version:
> > $ sudo alternatives --config java
> > (pick the one that points to Java 1.7)
> >
> > Somehow the JAVA_HOME got messed up after the first-ever execution of
> sqlline. So you need to:
> > $ JAVA_HOME=/usr/bin/java
> >
> > That's all there is to it.
> > $ cd /home/ec2-user/apache-drill-1.0.0-m1
> > $ bin/sqlline
> >
> >
> > Installation log … Sort of:
> > $ cd /home/ec2-user
> >
> > $ sudo yum install java-1.7.0-openjdk-devel
> > (I'm not sure if JDK is needed or just the JRE, so I put the JDK. This
> can also be the source of the java problems above…)
> >
> > $ curl
> http://people.apache.org/~jacques/apache-drill-1.0.0-m1.rc3/apache-drill-1.0.0-m1-binary-release.tar.gz> apache-drill-1.0.0-m1-binary-release.tar.gz
> >
> > $ tar -zxf apache-drill-1.0.0-m1-binary-release.tar.gz
> >
> > Cheers,
> > -Harri
> >
> >
> >
> >
> >
>
>
Re: AWS AMI for Drill M1
Posted by Michael Hausenblas <mi...@gmail.com>.
Awesome! One thing less on my to do list :)
Harri, how about writing a short blog post about it? I'm maintaining http://drill-user.org/ and this would make a great post, there.
Also, if anyone else here on the mailing list wants to use this blog to report on progress and/or experiences with Drill, please let me know!
Cheers,
Michael
--
Michael Hausenblas
Ireland, Europe
http://mhausenblas.info/
On 24 Sep 2013, at 21:10, Harri Kinnunen <Ha...@hitsaamo.fi> wrote:
> All,
>
> talked with Tim on the hangout about setting up a standalone instance of drill and publishing that as an Amazon AMI. The purpose of this is just to experiment with AWS so that Tim can carry forward with the AWS cluster setup. If I can later clear up the JAVA_HOME thing mentioned below, we could point to it as an alternative for a easy environment for trying Drill out.
>
> So ...
>
> The AMI is "ami-1343167a" and its name is "Drill M1 standalone", but you should find it using the ID in "EC2 / Images / AMIs". I launched it as t1.micro instance, and at least it was able to start up sqlline. Amazon says t1.micro has 615mb RAM.
>
>
> Running it:
> Make sure you're pointing to the correct Java version:
> $ sudo alternatives --config java
> (pick the one that points to Java 1.7)
>
> Somehow the JAVA_HOME got messed up after the first-ever execution of sqlline. So you need to:
> $ JAVA_HOME=/usr/bin/java
>
> That's all there is to it.
> $ cd /home/ec2-user/apache-drill-1.0.0-m1
> $ bin/sqlline
>
>
> Installation log … Sort of:
> $ cd /home/ec2-user
>
> $ sudo yum install java-1.7.0-openjdk-devel
> (I'm not sure if JDK is needed or just the JRE, so I put the JDK. This can also be the source of the java problems above…)
>
> $ curl http://people.apache.org/~jacques/apache-drill-1.0.0-m1.rc3/apache-drill-1.0.0-m1-binary-release.tar.gz > apache-drill-1.0.0-m1-binary-release.tar.gz
>
> $ tar -zxf apache-drill-1.0.0-m1-binary-release.tar.gz
>
> Cheers,
> -Harri
>
>
>
>
>