Reading from HDFS via HTTPFS using Hadoop Connector in Input Data Tool - 2019.1

[AnupD]

Hello,

I'm trying to read files from HDFS environment via HTTPFS using the built-in Hadoop connector in Input Data tool. I've the Kerberos set up as SSPI. I'm able to read (and write) from my local 2019.1 Designer (using my windows creds). I'm trying to test this connection on Alteryx Gallery (also 2019.1) but I get errors while publishing and running the workflow in Gallery.

I've set up Workflow Credentials to run the workflow as a Service Account in Gallery. That Service Account has necessary "run-as" permissions on Alteryx Server and has read/write access to HDFS. I get following errors when I try to publish/run. As I understand, workflows are pulled from MongoDB at run time and stored temporarily in XProcessCache folder in Service\Staging directory. Not sure what the process is trying to find in XProcessCache folder, when it tries to run the workflow. We've had a couple of users facing this issue, not just me. Also, I'm not sure if this has anything to do with Kerberos.

Has anybody else faced this issue? Thanks.

[BTran]

Hi Anup,

Can you ensure that the service account you have set-up has access to all the directories listed here?

https://help.alteryx.com/server/2018.1/admin/Configuration/RunasPermissions.htm

Also please note that the sub-folders, including XProcessCache will inherit those permissions.

If that doesn't work, please report back and we can take a closer look at why your service account can't access/find that folder.

Thanks,

Brendan T, Alteryx Support

[AnupD]

@BTran  Thanks for your response. Yes the the service account has all the required run-as permissions. I'm able to set up that account as a workflow credential (so I'm assuming all the required permissions are in place). But I'll double check and get back to you. Thanks.