You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@nifi.apache.org by "Mohammed Nadeem (Jira)" <ji...@apache.org> on 2020/07/17 19:46:00 UTC
[jira] [Resolved] (NIFI-7559) NiFi Cluster page shows multiple
entries for same node identity
[ https://issues.apache.org/jira/browse/NIFI-7559?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Mohammed Nadeem resolved NIFI-7559.
-----------------------------------
Resolution: Not A Problem
This behavior is seen if Statefulset pods are not mounted to PVC (ephemeral). So, every time a pod is restarted without preserving its state. NiFi clustering would think its a new node though it same node that was part of it before. Mounting nodes state will solve this issue
> NiFi Cluster page shows multiple entries for same node identity
> ---------------------------------------------------------------
>
> Key: NIFI-7559
> URL: https://issues.apache.org/jira/browse/NIFI-7559
> Project: Apache NiFi
> Issue Type: Bug
> Components: Core Framework, Core UI
> Affects Versions: 1.9.1, 1.9.2
> Environment: Linux
> Reporter: Mohammed Nadeem
> Priority: Major
> Attachments: image-2020-06-19-02-52-35-420.png
>
>
> In a containerized environment, When a NiFi node doesn't meet health probes performed by Kubelet (Kubernetes), then the node is killed and restarted. This may loop continuously causing *NiFi Cluster page* ( Hamburger Menu -> Cluster ) to show multiple entries for the same node that gets restarted again and again.
> To replicate the issue:
> 1. Setup single-node cluster ( statefulset application with replicaCount = 1)
> 2. Change configured probes such as Startup or Liveness to duration less than regular node start up time .
> 3. Scale new node (replicaCount), for example from 1 to 2. Then new scaled node *node-1* gets killed and restarted continuously by kubelet .
> Please check the screenshot below.
> *nifi-1:8080* is being added 7 times when it should be added just once, This shows incorrect *1/8* nodes cluster summary when expected behavior is just 1/2 nodes .
> !image-2020-06-19-02-50-27-674.png!
--
This message was sent by Atlassian Jira
(v8.3.4#803005)