Deletion of Files located in /user/hue/

[#1]

Good day,

We are in the process of tackling small file issues compounded in a variety of areas in HDFS.  One of these areas is "/user/hue/" (which currently has over 2000 small files in it).  I'm wondering what are the impacts of clearing this directory?  I was under the impression that everything "Hue dependent" was stored in the Hue database, but I also want to ensure that I don't break anything (specifically in /user/hue/oozie, but any "break" is a bad break).

Any advice is greatly appreciated.

Can't seem to locate any documentation on this.

Thanks,

Greg Frair

[#2]

Hi @gfrair ,

Unfortunately we don't have the any function for removing the
/user/hue/oozie/workspaces/hue-oozie-xxxxx.xxx so you have to manually remove them. 

Important: please make sure that the job related with the folder finishes already. Don't delete the folders when the job related with the folder is still running.

Thanks and hope this helps,

Li

[#3]

Thanks Li,

Can you provide some additional information?  If I was to delete all files in this folder over 7 days, should I not cause any issues?  Do these workspaces only get created by oozie when a job is about to start?  What about re-occurring jobs, do any rely on these source files?

Any assistance is greatly appreciated.

Thanks,

Greg

[#4]

Hi @gfrair ,

I am reaching out to our Hue expert in house and will get back to you with the answer soon.

Thanks!

Li

[#5]

Thanks very much!  Looking forward to the feedback 🙂

Greg

[#6]

For clarification as well, there are two directories within "/user/hue/oozie":

- deployments

- workspaces

I'm thinking deleting files within each of these will have their own set of implications.

Thanks again for looking into this!

Greg

[#7]

Hi @gfrair ,

Just to assure you I have not forgotten about this. I will continue monitoring this and providing feedback once I get some answers from Hue expert.

Thanks,

Li

[#8]

Thanks @lwang, Your diligence is greatly appreciated!

Looking forward to a response as this folder is growing quite out of control.

Greg Frair