Hi
I was supposed to work on getting two aarch64 VM systems operational for
bigtop CI tasks. But how to get such stuff working without finding out
how it works first? And does it work?
So I did builds of 1.2.1 and master branches on x86-64 under Debian 9
'stretch' and reported bunch of issues.
Today I did a build of 1.2.1 on fresh Debian 8 'jessie' on aarch64
system as this distribution was supported both by Linaro and by Bigtop
1.2.1 release.
Packages which built fine:
- alluxio
- apex
- bigtop-groovy
- bigtop-utils
- crunch
- flink
- flume
- giraph
- hadoop
- hama
- hive
- hue
- ignite-hadoop
- mahout
- oozie
- phoenix
- pig
- qfs
- solr
- spark1
- sqoop
- sqoop2
- tajo
- ycsb
- zeppelin
- zookeeper
Failures:
- ambari (uses x86-64 nodejs binaries)
- bigtop-jsvc ("Java Home not defined")
- bigtop-tomcat (have to check)
- datafu (no gradle binary)
- hbase (something about org.asciidoctor)
- kafka (no gradle binary)
- spark (no JAVA_HOME)
- tez (x86-64 nodejs binaries)
I followed "how to build bigtop trunk" document (using 'release-1.2.1'
git tag as base) on Debian 8 'jessie' system with 'jessie-backports'
repository enabled.
Will take a look at failures. Also plan to do Bigtop 'master' build on
fresh Debian 9 'stretch' aarch64 system to catch some bugs and to check
existing fixes.
Sorry for short mail without many details but it is 22:00 here and I do
that only because Olaf is online and updates Jira so my MiBand 2 thing
interrupted watching some TV series ;D
Ask, comment - will reply with details and provide logs.
