-
Notifications
You must be signed in to change notification settings - Fork 8
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Quickinstall now modular for upgrading, extra hints for Accumulo happiness #1
base: master
Are you sure you want to change the base?
Conversation
Upgrade Hadoop and Accumulo.
Less crowded inclusion in quickinstall-env.
Hey @dhutchis, thanks for the pull request. Looks pretty good. Couple of questions
I also had a thought it would be useful to be able to build multiple versions of the quickinstall from the same project. If we used profiles to manage Accumulo and Hadoop versions and kept directories separate, we could build a quickinstall for 1.6.2, 1.6.3 or 1.7.0. What are your thoughts on that? Thanks Mike |
Hey Mike, Yes I ran into a Hadoop linking error with later versions. Rather than try and sort it out, I kept Hadoop at the old version. It was more important to me to upgrade Accumulo. Regarding profiles for the multiple versions, I think it is sufficient enough to override the pom variables on the command line. Anyone who does that is not a novice. They also have to override the SHA checksum. I forgot about the javadocs. How do you generate them? Do you have a script to copy them over from the right places after running "mvn site" on accumulo source? An alternative to maintaining the manual and javadocs in the repo is to redirect the user to the Accumulo javadocs online and make commands like accumulo-manual and accumulo-javadoc. The URLs are formed from the major and minor version number of Accumulo. Probably similar for Hadoop and Zookeeper. On the other hand, having local copies is kinda cool. |
Thanks @dhutchis. I will look this over this weekend. Quick glance looked good. Not sure the best way to maintain multiple versions. I think I am going to make a 1.6.1 branch for current master and a 1.7.1 branch for this. Maybe even remove the master branch and just keep the branches separate. |
It is pretty easy to change versions. All the user needs to do is change Because we tied If it were me in charge, I would maintain master at the latest version and On Fri, Apr 8, 2016 at 7:11 PM, Michael Wall [email protected]
|
The Accumulo version is easy, rebuilding the docs take a minute. I was worried about rebuilding libhadoop stuff to support different Hadoop versions. I like your suggestion of making master the latest version of what has been built and just branching before supporting a new version. Thanks |
Hey @mjwall, did this get merged in? |
ugh, no. Sorry Dylan. I pulled it down and made one tweak, but wanted to Mike On Wed, Jun 8, 2016 at 10:35 PM, Dylan Hutchison [email protected]
|
These changes upgrade quickinstall to Accumulo 1.7.0 and pave the way for future upgrades. I pulled the version hardcoded constants to reference pom variables and used resource filtering to place the constants into the scripts. An ant task renames the hadoop/zookeeper/accumulo directories to the correct version.
Included extra hints in the README for setting up ssh and for resolving common warnings generated by Accumulo, namely too high vm swappiness and too low max open file limit. Also fixed the hadoop instance.volumes warning.