Hi,

Is it possible to run multiple data nodes on a single machine? I
currently have a machine with multiple disks and enough disk capacity
for replication across them. I don't need redundancy at the machine
level but would like to be able to handle a single disk failure.

So I was thinking if I can run multiple DataNodes on a single machine
each assigned a separate disk that would give me the protection I need
against disk failure.

Can anyone give me any insights in to how I would setup multiple
DataNodes to run on a single machine? Thanks in advance,

Cheers Arv

Hi, Arv,

Actually, several days ago, I deployed a system which is similar with your
requirements

In our cluster environment, since I have to run modified hadoop, we invoked
two namenodes, two jobtrackers, two trackers on each node, and as you
mentioned, two datanodes in single host,

What you have to do is that you should ensure there is no conflicts on the
port of datanodes occupy(modify the configuration files), and for assigning
separate disks , just set the different root directory of your HDFS, I guess
that you also have to run two namenodes just like we have done

Cheers

Nan

On Wed, Sep 15, 2010 at 11:50 PM, Arv Mistry <[EMAIL PROTECTED]> wrote:

> Hi,
>
> Is it possible to run multiple data nodes on a single machine? I
> currently have a machine with multiple disks and enough disk capacity
> for replication across them. I don't need redundancy at the machine
> level but would like to be able to handle a single disk failure.
>
> So I was thinking if I can run multiple DataNodes on a single machine
> each assigned a separate disk that would give me the protection I need
> against disk failure.
>
> Can anyone give me any insights in to how I would setup multiple
> DataNodes to run on a single machine? Thanks in advance,
>
> Cheers Arv
>

On 09/15/2010 11:50 AM, Arv Mistry wrote:
> Hi,
>
> Is it possible to run multiple data nodes on a single machine? I
> currently have a machine with multiple disks and enough disk capacity
> for replication across them. I don't need redundancy at the machine
> level but would like to be able to handle a single disk failure.
>
> So I was thinking if I can run multiple DataNodes on a single machine
> each assigned a separate disk that would give me the protection I need
> against disk failure.
>
> Can anyone give me any insights in to how I would setup multiple
> DataNodes to run on a single machine? Thanks in advance,
>
> Cheers Arv

I guess you *could*, but it doesn't seem to me it would make much sense
in a production environment, since the 2 data nodes running on the
machine would be competing with each other for CPU usage, network
bandwidth, etc.

DR

Hello Arv,
It is possible to run multiple datanodes on a single machine, and this can be useful for small-scale test scenarios.  Also you mentioned in your previous message that you have a Hadoop implementation with only one physical datanode server and want to replicate within it, between spindles.  This also makes sense, and will work.  Of course, if you have two datanodes running you will get only order-2 replication, not order-3, even if the replication has been set to 3.

I will describe the config in a moment, but I would first like to point out that in clusters with even a few datanode servers, one is better off with cross-server replication.  Without cross-server replication, losing the System disk will make ALL data volumes unavailable.  And of course, multiple datanodes running on one server will compete for cores, NICs, bus, and memory access, even if not for spindles.

A previous responder suggested running two namenodes also, but it wasn't clear whether he meant two primaries or one primary and one secondary/checkpoint nameserver.  The latter is fine, but running two primary namenodes is definitely not the thing to do!

Anyway, here's how you set it up.  I have done this recently with v0.21.0, with two datanode processes in a single box (along with namenode sharing the same box), and it did replicate correctly between the two.  I haven't tried it with > 2 datanodes, and I don't know what the impact on process efficiency would be, but that would probably work too.

1. In your HADOOP_HOME directory, copy the "conf" directory to, say, "conf2".

2. In the conf2 directory, edit as follows:

  a) In hadoop-env.sh, provide unique non-default HADOOP_IDENT_STRING, e.g. ${USER}_02
  b) In hdfs-site.xml, change dfs.data.dir to show the desired targets/volumes for datanode#2, and of course make sure the corresponding target directories exist.  Also remove these targets from the dfs.data.dir target list for datanode#1 in conf/hdfs-site.xml.
  c) in hdfs-site.xml, set the four following "address:port" strings to something non-conflicting with the other datanode and other processes running on this box:
    - dfs.datanode.address  (default 0.0.0.0:50010)
    - dfs.datanode.ipc.address  (default 0.0.0.0:50020)
    - dfs.datanode.http.address  (default 0.0.0.0:50075)
    - dfs.datanode.https.address  (default 0.0.0.0:50475)
Note: the defaults above are what datanode#1 is probably running on.  I added 2 to each port number for datanode#2 and it seemed to work okay.  You might also wish to note the default ports associated with the namenode and job/task tracker processes, in case they are running on the same box:
    - fs.default.name  0.0.0.0:9000
    - dfs.http.address  0.0.0.0:50070
    - dfs.https.address  0.0.0.0:50470
    - dfs.secondary.http.address  0.0.0.0:50090
    - mapred.job.tracker.http.address  0.0.0.0:50030
    - mapred.task.tracker.report.address  127.0.0.1:0
    - mapred.task.tracker.http.address  0.0.0.0:50060

3. At this point, launching with:
    bin/hdfs --config $HADOOP_HOME/conf2 datanode
will work.  To make it convenient to launch as a service, you can add a couple lines to the end of the bin/start-dfs.sh script like:
    HADOOP_CONF_DIR2=$HADOOP_HOME/conf2
    "$HADOOP_COMMON_HOME"/bin/hadoop-daemons.sh --config $HADOOP_CONF_DIR2 --script "$bin"/hdfs start datanode $dataStartOpt

Hope this helps,
--Matt

On Sep 15, 2010, at 8:50 AM, Arv Mistry wrote:

Hi,

Is it possible to run multiple data nodes on a single machine? I
currently have a machine with multiple disks and enough disk capacity
for replication across them. I don't need redundancy at the machine
level but would like to be able to handle a single disk failure.

So I was thinking if I can run multiple DataNodes on a single machine
each assigned a separate disk that would give me the protection I need
against disk failure.

Can anyone give me any insights in to how I would setup multiple
DataNodes to run on a single machine? Thanks in advance,

Cheers Arv

to me, the notion of replication is to provide a failsafe mechanism in case
some nodes go down, hence running two datanode on a single host does not
serve this basic purpose. As already mentioned you can run two data nodes
using two instanced of hadoop with different ports. In case you do not want
to replicate , you can simply set the replication factor to 1.

On Thu, Sep 16, 2010 at 6:15 AM, Matthew Foley <[EMAIL PROTECTED]> wrote:

> Hello Arv,
> It is possible to run multiple datanodes on a single machine, and this can
> be useful for small-scale test scenarios.  Also you mentioned in your
> previous message that you have a Hadoop implementation with only one
> physical datanode server and want to replicate within it, between spindles.
>  This also makes sense, and will work.  Of course, if you have two datanodes
> running you will get only order-2 replication, not order-3, even if the
> replication has been set to 3.
>
> I will describe the config in a moment, but I would first like to point out
> that in clusters with even a few datanode servers, one is better off with
> cross-server replication.  Without cross-server replication, losing the
> System disk will make ALL data volumes unavailable.  And of course, multiple
> datanodes running on one server will compete for cores, NICs, bus, and
> memory access, even if not for spindles.
>
> A previous responder suggested running two namenodes also, but it wasn't
> clear whether he meant two primaries or one primary and one
> secondary/checkpoint nameserver.  The latter is fine, but running two
> primary namenodes is definitely not the thing to do!
>
> Anyway, here's how you set it up.  I have done this recently with v0.21.0,
> with two datanode processes in a single box (along with namenode sharing the
> same box), and it did replicate correctly between the two.  I haven't tried
> it with > 2 datanodes, and I don't know what the impact on process
> efficiency would be, but that would probably work too.
>
> 1. In your HADOOP_HOME directory, copy the "conf" directory to, say,
> "conf2".
>
> 2. In the conf2 directory, edit as follows:
>
>  a) In hadoop-env.sh, provide unique non-default HADOOP_IDENT_STRING, e.g.
> ${USER}_02
>  b) In hdfs-site.xml, change dfs.data.dir to show the desired
> targets/volumes for datanode#2, and of course make sure the corresponding
> target directories exist.  Also remove these targets from the dfs.data.dir
> target list for datanode#1 in conf/hdfs-site.xml.
>  c) in hdfs-site.xml, set the four following "address:port" strings to
> something non-conflicting with the other datanode and other processes
> running on this box:
>    - dfs.datanode.address  (default 0.0.0.0:50010)
>    - dfs.datanode.ipc.address  (default 0.0.0.0:50020)
>    - dfs.datanode.http.address  (default 0.0.0.0:50075)
>    - dfs.datanode.https.address  (default 0.0.0.0:50475)
> Note: the defaults above are what datanode#1 is probably running on.  I
> added 2 to each port number for datanode#2 and it seemed to work okay.  You
> might also wish to note the default ports associated with the namenode and
> job/task tracker processes, in case they are running on the same box:
>    - fs.default.name  0.0.0.0:9000
>    - dfs.http.address  0.0.0.0:50070
>    - dfs.https.address  0.0.0.0:50470
>    - dfs.secondary.http.address  0.0.0.0:50090
>    - mapred.job.tracker.http.address  0.0.0.0:50030
>    - mapred.task.tracker.report.address  127.0.0.1:0
>    - mapred.task.tracker.http.address  0.0.0.0:50060
>
> 3. At this point, launching with:
>    bin/hdfs --config $HADOOP_HOME/conf2 datanode
> will work.  To make it convenient to launch as a service, you can add a
> couple lines to the end of the bin/start-dfs.sh script like:
>    HADOOP_CONF_DIR2=$HADOOP_HOME/conf2
>    "$HADOOP_COMMON_HOME"/bin/hadoop-daemons.sh --config $HADOOP_CONF_DIR2
> --script "$bin"/hdfs start datanode $dataStartOpt
>
> Hope this helps,
> --Matt
>
> On Sep 15, 2010, at 8:50 AM, Arv Mistry wrote:
>
> Hi,
>
> Is it possible to run multiple data nodes on a single machine? I

Thanks for the responses, I especially appreciate the details Matthew!

Just for the record, I appreciate that having multiple DataNodes on a
single machine defeats the purpose or the advantages given by having
them spread across machines across racks. I intend to go to that model
as we grow.

Cheers Arv
-----Original Message-----
From: Matthew Foley [mailto:[EMAIL PROTECTED]]
Sent: September 15, 2010 8:45 PM
To: [EMAIL PROTECTED]
Cc: Matthew Foley
Subject: Re: Multiple DataNodes on a single machine

Hello Arv,
It is possible to run multiple datanodes on a single machine, and this
can be useful for small-scale test scenarios.  Also you mentioned in
your previous message that you have a Hadoop implementation with only
one physical datanode server and want to replicate within it, between
spindles.  This also makes sense, and will work.  Of course, if you have
two datanodes running you will get only order-2 replication, not
order-3, even if the replication has been set to 3.

I will describe the config in a moment, but I would first like to point
out that in clusters with even a few datanode servers, one is better off
with cross-server replication.  Without cross-server replication, losing
the System disk will make ALL data volumes unavailable.  And of course,
multiple datanodes running on one server will compete for cores, NICs,
bus, and memory access, even if not for spindles.

A previous responder suggested running two namenodes also, but it wasn't
clear whether he meant two primaries or one primary and one
secondary/checkpoint nameserver.  The latter is fine, but running two
primary namenodes is definitely not the thing to do!
Anyway, here's how you set it up.  I have done this recently with
v0.21.0, with two datanode processes in a single box (along with
namenode sharing the same box), and it did replicate correctly between
the two.  I haven't tried it with > 2 datanodes, and I don't know what
the impact on process efficiency would be, but that would probably work
too.

1. In your HADOOP_HOME directory, copy the "conf" directory to, say,
"conf2".

2. In the conf2 directory, edit as follows:

  a) In hadoop-env.sh, provide unique non-default HADOOP_IDENT_STRING,
e.g. ${USER}_02
  b) In hdfs-site.xml, change dfs.data.dir to show the desired
targets/volumes for datanode#2, and of course make sure the
corresponding target directories exist.  Also remove these targets from
the dfs.data.dir target list for datanode#1 in conf/hdfs-site.xml.
  c) in hdfs-site.xml, set the four following "address:port" strings to
something non-conflicting with the other datanode and other processes
running on this box:
    - dfs.datanode.address  (default 0.0.0.0:50010)
    - dfs.datanode.ipc.address  (default 0.0.0.0:50020)
    - dfs.datanode.http.address  (default 0.0.0.0:50075)
    - dfs.datanode.https.address  (default 0.0.0.0:50475)
Note: the defaults above are what datanode#1 is probably running on.  I
added 2 to each port number for datanode#2 and it seemed to work okay.
You might also wish to note the default ports associated with the
namenode and job/task tracker processes, in case they are running on the
same box:
    - fs.default.name  0.0.0.0:9000
    - dfs.http.address  0.0.0.0:50070
    - dfs.https.address  0.0.0.0:50470
    - dfs.secondary.http.address  0.0.0.0:50090
    - mapred.job.tracker.http.address  0.0.0.0:50030
    - mapred.task.tracker.report.address  127.0.0.1:0
    - mapred.task.tracker.http.address  0.0.0.0:50060

3. At this point, launching with:
    bin/hdfs --config $HADOOP_HOME/conf2 datanode
will work.  To make it convenient to launch as a service, you can add a
couple lines to the end of the bin/start-dfs.sh script like:
    HADOOP_CONF_DIR2=$HADOOP_HOME/conf2
    "$HADOOP_COMMON_HOME"/bin/hadoop-daemons.sh --config
$HADOOP_CONF_DIR2 --script "$bin"/hdfs start datanode $dataStartOpt

Hope this helps,
--Matt

On Sep 15, 2010, at 8:50 AM, Arv Mistry wrote:

Hi,

Is it possible to run multiple data nodes on a single machine? I
currently have a machine with multiple disks and enough disk capacity
for replication across them. I don't need redundancy at the machine
level but would like to be able to handle a single disk failure.

So I was thinking if I can run multiple DataNodes on a single machine
each assigned a separate disk that would give me the protection I need
against disk failure.

Can anyone give me any insights in to how I would setup multiple
DataNodes to run on a single machine? Thanks in advance,

Cheers Arv