HDP on Linux – Installation Forum
Just and FYI,
I’m doing some eval work with hadoop and noted some things in your documentation you might want to take a look at.
The systems I’m working on in my lab and not accessible to/from the outside so I attempted to use your reference document bk-reference-20140210.pdf. For my situation I need to deploy behind a firewall with temporary access to the internet so I’m setting up local repos. I’m using CentOS6.5 currently for now. This brings us to section 18.104.22.168. Using table 4.6 I used wget to fetch the hdp.repo file for yum to use. The problem is that your recommend arguments, specifically HDP and HDP-22.214.171.124, for reposync command that follow are not matched in the hdp.repo file I just downloaded so they fail.
I poked around and found what I thought were the correct names (HDP-2.x and Updates-HDP-2.x) and tried wget on those but I end up with 126.96.36.199 (from centos6/2.x/GA) so I gave up and just fetched the whole HDP-188.8.131.52-centos6-rpm.tar.gz from the HDP/centos6/ directory. The utils directory seems to work from above and seems current so I’ll run with it.
The directory names used in the createrepo reflect the reposync names. I haven’t gotten to the Ambari installation docs to see what names are referenced there but you should probably check they match correctly too.
Also noted that ambari repo has similar problems too and it’s repo files reflect a down rev so I bypassed that too. I did try to use a label that matched the repo file but ultimately killed the process as it was attempting to mirror atrpms – all 2600+ which is a little out of scope for my project at the moment. This is likely my error but it might be helpful.
I did search the forums and saw several threads that I suspect may be related to the hdp.repo being out of sync with the current directory structure which may be an underlying problem.
Anyway, hope this is helpful and I’m not missing the obvious.
Support from the Experts
A HDP Support Subscription connects you experts with deep experience running Apache Hadoop in production, at-scale on the most demanding workloads.
Become HDP Certified
Real world training designed by the core architects of Hadoop. Scenario-based training courses are available in-classroom or online from anywhere in the world