It would be really helpful if you can send in your spec that you used and
has got issues.

I am gonna give it a try myself also with all the disks as the mounted
volumes to see if issue exists.

On Mon, Nov 9, 2020 at 4:11 AM muhyid72 <[email protected]> wrote:

> Hi Sushil,
> Thanks for support
> I am trying to run standalone NiFi 1.12.1 (not cluster) on Kubernetes. I am
> planning to migrate my NiFi instance from VM to Kubernetes.
> I deployed official image to Kubernetes (standalone one replica, not
> cluster). I used Persistent Volume and Persistent Volume Claim for
> persistency requirements.
> I tried that scenario on my virtual test environment (with NFS) and Azure
> AKS (with standard Azure Disk):
> I mounted all folders which are required for persistency (conf, state,
> flowfile_repository etc.) one by one with PersistentVolumeClaim on the pod
> (I used only one pod).
> I didn't encounter any problem (I saw related files in the mounted folders,
> they were successfully redirected to mounted disks) till try to conf
> folder.
> When I mount conf folder to Kubernetes (on my virtual test environment and
> Azure AKS ) through volumeMounts in the yaml file I saw those errors:
> "replacing target file  /opt/nifi/nifi-current/conf/nifi.properties"
> "sed: can't read /opt/nifi/nifi-current/conf/nifi.properties: No such file
> or directory"
> It seems when container creates it doesn't see the mounted volume for conf
> folder (nifi.properties file) because nifi.properties file is creating by
> application before VolumeMounts
> When I investigated more deeply I encountered that bug record Apache
> NiFi / NIFI-6484 (https://issues.apache.org/jira/browse/NIFI-6484)
>
> Shortly I would like to use persistent disk for not losing any important
> data (nifi.properties, users.xml, authorizations.xml etc.). I did all
> persistency required folders with VolumeMount and PersistentVolumeClaim
> except the conf folder
>
> My PoC Environment:
> Apache Nifi 1.12.1 Official Image (https://hub.docker.com/r/apache/nifi/)
> 1 master 2 worker nodes (on VirtualBox and on Azure Kubernetes Services
> (AKS))
> Deployment is standalone not cluster therefore I am using standard
> deployment yaml with single replica.
>
> Kind Regards
>
>
>
> --
> Sent from: http://apache-nifi-users-list.2361937.n4.nabble.com/
>

Reply via email to