You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@nifi.apache.org by "Andrew Grande (JIRA)" <ji...@apache.org> on 2015/11/18 21:37:11 UTC

[jira] [Created] (NIFI-1190) 2 NICs - can't have a cluster node serve both the web ui and process requests

Andrew Grande created NIFI-1190:
-----------------------------------

             Summary: 2 NICs - can't have a cluster node serve both the web ui and process requests
                 Key: NIFI-1190
                 URL: https://issues.apache.org/jira/browse/NIFI-1190
             Project: Apache NiFi
          Issue Type: Bug
          Components: Core Framework
    Affects Versions: 0.4.0
            Reporter: Andrew Grande


In a NiFi cluster with 2 network interfaces I was unable to achieve the following:
* Node registering with NCM and processing successfully
* Node being accessible via the Web host/port on a public IP

The problem seems to come from a requirement to specify *nifi.web.http.host* for the node. See a comment here https://github.com/aperepel/docker-nifi/blob/master/docker/run.sh#L28

Basically, if I want it to bind to all interfaces (and not only internal IP, which is not accessible to the user browser), I have to specify 0.0.0.0. In a dynamic environment I don't know the ip at deploy time yet. But, this 0.0.0.0 address is then passed as is to the NCM and it can't talk back to the node.

If I specify a hostname, it resolves to an internal IP, and, while NCM and the node are happy, the user can't access the Web/API port for the node (NCM works).

We should potentially split these 2 concerns into 2 properties:
* Web HTTP/HTTPS host
* Cluster protocol chat



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)