You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@nifi.apache.org by "Firenz (JIRA)" <ji...@apache.org> on 2019/01/18 16:31:00 UTC

[jira] [Commented] (NIFI-5964) InvokeHTTP, Dynamic Headers not sent (Cluster)

    [ https://issues.apache.org/jira/browse/NIFI-5964?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16746454#comment-16746454 ] 

Firenz commented on NIFI-5964:
------------------------------

Additionnal infos : it happened with two differents invokeHttp "flows", requesting differents URLs. The shared property of they URLs is read timeout issues (slow APIs).

No issues when using NIFI 1.4.0 (but we had no cluster).

> InvokeHTTP, Dynamic Headers not sent (Cluster)
> ----------------------------------------------
>
>                 Key: NIFI-5964
>                 URL: https://issues.apache.org/jira/browse/NIFI-5964
>             Project: Apache NiFi
>          Issue Type: Bug
>    Affects Versions: 1.8.0
>         Environment: Oracle Jre1.8.0_181
> CentOS Linux release 7.5.1804 (Core)
> 3.10.0-862.11.6.el7.x86_64
> Zookeeper 3.4.12 (same VM, dedicated to NIFI, not the standalone of NIFI)
>            Reporter: Firenz
>            Priority: Critical
>         Attachments: Step 1 - Bulletin.png, Step 1 - InvokeHTTP conf.png, Step 2 - Bulletin.png, Step 2 - InvokeHTTP.png, Step 3 - Bulletin.png
>
>
> InvokeHTTP does not always send HTTP Headers specified with dynamic Properties.
>  
> This issue has only been observed with a 3 nodes cluster (3 NIFI, 3 zookeeper, 3 VMs). This works fine on standalone NIFI (our others stagings). Our flow is not connection-load-balanced. The first processor is "PRIMARY" tagged.
>  
> Step 0 
> Everything works. We do some STOP, START. The InvokeHTTP is not modified by anyone.
> Step 1
> We observe the HTTP Headers are not sent anymore. The flow files on the 3 servers are identicals unzipped (cksum). The flow run on our PRIMARY node "014"
> Step 2
> We add manually a new dynamic header in the InvokeHTTP (x-test). (With our node "019", our https UI is loadbalanced) 
> The new dymanic header is sent, not the initial ones.
> Step 3
> We offload the PRIMARY node. The traffic goes to a new PRIMARY node ("005"). It works : the HTTP headers are sent.
>  
> Some pictures provided for each step.
>  
> Note : the headers "masked in the snapshots"  are string of 36 chars max (UUID/secrets)
> So, depending on the node, the HTTP Header are not sent.
>  
>  
>  
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)