Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Accumulo, mail # dev - Proper way to build an accumulo RPM?


Copy link to this message
-
Proper way to build an accumulo RPM?
Drew Farris 2012-10-19, 16:59
Hi All,

Given the impending 1.4.2 release, I'm curious, what are the proper
steps from a clean checkout to build an RPM?

I tried to do a rpm build last night, and while what I built was able
to install, be configured and start, it was significantly smaller than
the 1.4.1 release rpm, was missing some of the example configuration
files, among other things.

The rpm initially built with amd64 as the architecutre string,
Centos/RHEL prefer x86_64. The amd64 rpm installs fine with rpm
--install, but it confuses the heck out of yum when deployed to a
repository.

I managed to tweak the maven rpm plugin configuration sufficiently to
produce an rpm with the proper architecture, but as described above it
seemed to be missing things.

I understand that the rpm build process is currently less than ideal,
I'm just trying to see how far I can get with the current process with
minor tweaks instead of revamping it which is beyond what I have the
resources to do at this point.

IIRC, the steps I used were:

mvn package && mvn assembly:single -N
( cd ./src/server/src/main/c++ ; make )
mvn rpm:rpm

The last step did not complete successfully as reactor attempted to
execute rpm:rpm for all sub-modules and failed at start.

I performed this all on Centos 6.3 x86_64 -- I had to tweak the native
Makefiles to no build the 32 bit libraries because I had not installed
the 32bit toolchain.

Thanks,

Drew