[
https://issues.apache.org/jira/browse/NIFI-8100?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
David Handermann resolved NIFI-8100.
------------------------------------
Assignee: David Handermann
Resolution: Resolved
Thanks for the comment [~larsfrancke], yes, this is effectively implemented in
the Management Server health endpoints.
As you mentioned, the default configuration limits the Management Server to
listen on 127.0.0.1. This can be changed to listen on a different address,
using the
[management.server.address|https://nifi.apache.org/nifi-docs/administration-guide.html#bootstrap_properties]
property in bootstrap.conf. As noted in the project [Security
Model|https://nifi.apache.org/documentation/security/] changing the listen
address has security implications that are the responsibility of the
administrator to control.
> Provide a /health endpoint that do not require authentication for use in
> LoadBalancer (nifi UI)
> -----------------------------------------------------------------------------------------------
>
> Key: NIFI-8100
> URL: https://issues.apache.org/jira/browse/NIFI-8100
> Project: Apache NiFi
> Issue Type: Improvement
> Reporter: Ruben Laguna
> Assignee: David Handermann
> Priority: Major
>
> Usually load balancers with AWS Application Load Balancer use periodic health
> checks to determine which node it should forward requests to.
> Today we have the /status api endpoint but that is not suitable because it
> really requires authentication on a secure nifi installation and there is no
> way to provide authorization headers in the health checks performed by load
> balancers.
> So I would like to have a /heath or /health-check endpoint that can be "open"
> that answers 200 OK *when the node is connected to the cluster and a 4xx-5xx
> error if the node is disconnected* ("Only a node connected to a cluster can
> process the request", "This node is currently not connect to the cluster. Any
> modifications to the data flow made here will not replicate across the
> cluster").
> This will simplify secure nifi deployment with load balancing, right now if a
> node gets disconnected users will be still be routed to it even if that node
> is really not usable.
> Here I'm referring to *Load balancer for the NiFi web UI*, and just checking
> for / or /nifi is not good enough because t he UI may be up (GET / 200OK) but
> not usable (the node is disconnected from the cluster).
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
