Network.publish_host for an address outside a docker container

Elastic Stack Elasticsearch
1

I am trying to do a setup like this:

machine1 - 192.168.0.10

9300 => 9300 Docker container elasticsearch1 internal vlan 172.17.0.68

Elasticsearch transport running on port 9300

network.publish_host 192.168.10:9300

9301 => 9300 Docker container elasticsearch2 internal vlan 172.17.0.69

Elasticsearch transport running on port 9300

network.publish_host 192.168.10:9301

machine2 - 192.168.0.20

9300 => 9300 Docker container elasticsearch1 internal vlan 172.17.0.68

Elasticsearch transport running on port 9300

network.publish_host 192.168.20:9300

9301 => 9300 Docker container elasticsearch2 internal vlan 172.17.0.69

Elasticsearch transport running on port 9300

network.publish_host 192.168.20:9301

So if I define machine1 / elasticsearch1 as my master, I should be able to
set unicast on the three other nodes as 192.168.0.10:9300, and as listed
above, the network.publish_host on each node, BUT
Each node in its docker container does not see the address it should be
publishing, and it seems that this setting not only says "publish this
address" but also "bind to this address".

Is there any way to publish an address which the node cannot actually bind
to to allow machines in this type of a constellation to connect to each
other?

Thank you!
-Robin-

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/ababbf50-ecae-4988-a4a0-af4ea2fb7412%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

2

Nope.
There are a few docker users on the IRC channel that might be able to give
you some assistance or pointer though.

On 19 November 2014 00:38, Robin Clarke robin13@gmail.com wrote:

I am trying to do a setup like this:

machine1 - 192.168.0.10

9300 => 9300 Docker container elasticsearch1 internal vlan 172.17.0.68

Elasticsearch transport running on port 9300

network.publish_host 192.168.10:9300

9301 => 9300 Docker container elasticsearch2 internal vlan 172.17.0.69

Elasticsearch transport running on port 9300

network.publish_host 192.168.10:9301

machine2 - 192.168.0.20

9300 => 9300 Docker container elasticsearch1 internal vlan 172.17.0.68

Elasticsearch transport running on port 9300

network.publish_host 192.168.20:9300

9301 => 9300 Docker container elasticsearch2 internal vlan 172.17.0.69

Elasticsearch transport running on port 9300

network.publish_host 192.168.20:9301

So if I define machine1 / elasticsearch1 as my master, I should be able to
set unicast on the three other nodes as 192.168.0.10:9300, and as listed
above, the network.publish_host on each node, BUT
Each node in its docker container does not see the address it should be
publishing, and it seems that this setting not only says "publish this
address" but also "bind to this address".

Is there any way to publish an address which the node cannot actually
bind to to allow machines in this type of a constellation to connect to
each other?

Thank you!
-Robin-

--
You received this message because you are subscribed to the Google Groups
"elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an
email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/elasticsearch/ababbf50-ecae-4988-a4a0-af4ea2fb7412%40googlegroups.com
https://groups.google.com/d/msgid/elasticsearch/ababbf50-ecae-4988-a4a0-af4ea2fb7412%40googlegroups.com?utm_medium=email&utm_source=footer
.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CAF3ZnZkE_CrKoL6brPMN6qcHE_LEMOvghMzjx5xkz8zgvqaWWg%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

3

You need to describe any network troubleshooting steps you've tried (like
telnet or curl commands) that verify network connectivity to open ports
with an active service responding.

It might also be helpful to understand how you're doing your network
configuration, particularly is you're using vlans (why? Why not a simple
network connection?) If you're really using vlans, you might need to
describe how those are set up if you can't configure network connectivity
correctly. The issue can be, if you've really set up vlans, you might need
to also describe the Bridge Control Device you've created.

You might also post the Dockerfile you've built (I assume you must, to
incorporate your customizations) which would provide a clearer picture as
well as a sample "run" command complete with the switches you're using.

Tony

On Tuesday, November 18, 2014 5:38:48 AM UTC-8, Robin Clarke wrote:

I am trying to do a setup like this:

machine1 - 192.168.0.10

9300 => 9300 Docker container elasticsearch1 internal vlan 172.17.0.68

Elasticsearch transport running on port 9300

network.publish_host 192.168.10:9300

9301 => 9300 Docker container elasticsearch2 internal vlan 172.17.0.69

Elasticsearch transport running on port 9300

network.publish_host 192.168.10:9301

machine2 - 192.168.0.20

9300 => 9300 Docker container elasticsearch1 internal vlan 172.17.0.68

Elasticsearch transport running on port 9300

network.publish_host 192.168.20:9300

9301 => 9300 Docker container elasticsearch2 internal vlan 172.17.0.69

Elasticsearch transport running on port 9300

network.publish_host 192.168.20:9301

So if I define machine1 / elasticsearch1 as my master, I should be able to
set unicast on the three other nodes as 192.168.0.10:9300, and as listed
above, the network.publish_host on each node, BUT
Each node in its docker container does not see the address it should be
publishing, and it seems that this setting not only says "publish this
address" but also "bind to this address".

Is there any way to publish an address which the node cannot actually
bind to to allow machines in this type of a constellation to connect to
each other?

Thank you!
-Robin-

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/6e13acb4-c998-4cd3-b917-516d1283dff2%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

4

In the we used docker plugin pipework-master to create a docker interface
mapped to an external interface, so now we have multiple nodes on a single
docker host communicating between multiple hosts using multicast.

-Robin-

On Tuesday, 18 November 2014 23:54:00 UTC+1, Tony Su wrote:

You need to describe any network troubleshooting steps you've tried (like
telnet or curl commands) that verify network connectivity to open ports
with an active service responding.

It might also be helpful to understand how you're doing your network
configuration, particularly is you're using vlans (why? Why not a simple
network connection?) If you're really using vlans, you might need to
describe how those are set up if you can't configure network connectivity
correctly. The issue can be, if you've really set up vlans, you might need
to also describe the Bridge Control Device you've created.

You might also post the Dockerfile you've built (I assume you must, to
incorporate your customizations) which would provide a clearer picture as
well as a sample "run" command complete with the switches you're using.

Tony

On Tuesday, November 18, 2014 5:38:48 AM UTC-8, Robin Clarke wrote:

I am trying to do a setup like this:

machine1 - 192.168.0.10

9300 => 9300 Docker container elasticsearch1 internal vlan 172.17.0.68

Elasticsearch transport running on port 9300

network.publish_host 192.168.10:9300

9301 => 9300 Docker container elasticsearch2 internal vlan 172.17.0.69

Elasticsearch transport running on port 9300

network.publish_host 192.168.10:9301

machine2 - 192.168.0.20

9300 => 9300 Docker container elasticsearch1 internal vlan 172.17.0.68

Elasticsearch transport running on port 9300

network.publish_host 192.168.20:9300

9301 => 9300 Docker container elasticsearch2 internal vlan 172.17.0.69

Elasticsearch transport running on port 9300

network.publish_host 192.168.20:9301

So if I define machine1 / elasticsearch1 as my master, I should be able
to set unicast on the three other nodes as 192.168.0.10:9300, and as
listed above, the network.publish_host on each node, BUT
Each node in its docker container does not see the address it should be
publishing, and it seems that this setting not only says "publish this
address" but also "bind to this address".

Is there any way to publish an address which the node cannot actually
bind to to allow machines in this type of a constellation to connect to
each other?

Thank you!
-Robin-

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/4c0c3d5d-d416-4c82-846d-4a88a974b7e7%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.