One way out of this dilemma is to release one build, 0.92-security Part of the justification is that the underlying hadoop distro is secure. The assumption is that 0.92-security build can adequately handle insecure hbase operations.
This would simplify the release process a lot. Cheers On Mon, Nov 21, 2011 at 4:44 PM, Ted Yu <yuzhih...@gmail.com> wrote: > Sending to wider audience. > > > On Mon, Nov 21, 2011 at 4:28 PM, Ted Yu <yuzhih...@gmail.com> wrote: > >> On TRUNK, without -P security, TestMasterFailover passed. >> With -P security, I got: >> >> Running org.apache.hadoop.hbase.master.TestMasterFailover >> >> Results : >> >> Tests run: 0, Failures: 0, Errors: 0, Skipped: 0 >> >> [INFO] >> ------------------------------------------------------------------------ >> [INFO] BUILD FAILURE >> [INFO] >> ------------------------------------------------------------------------ >> [INFO] Total time: 7:44.641s >> [INFO] Finished at: Mon Nov 21 15:57:39 PST 2011 >> [INFO] Final Memory: 27M/123M >> [INFO] >> ------------------------------------------------------------------------ >> [ERROR] Failed to execute goal >> org.apache.maven.plugins:maven-surefire-plugin:2.11-TRUNK-HBASE-2:test >> (default-test) on project hbase: No tests were executed! (Set >> -DfailIfNoTests=false to ignore this error.) -> [Help 1] >> [ERROR] >> >> On Mon, Nov 21, 2011 at 1:03 PM, Ted Yu <yuzhih...@gmail.com> wrote: >> >>> Hi, >>> I want to get your opinion on how we manage the complexity of >>> maintaining clean Jenkins builds. >>> We now have 6 builds, including >>> HBase-TRUNK-on-Hadoop-23<https://builds.apache.org/view/G-L/view/HBase/job/HBase-TRUNK-on-Hadoop-23/> >>> . >>> >>> There is one additional dimension after the two -security builds were >>> introduced. >>> Namely certain test(s) would fail in -security build >>> (org.apache.hadoop.hbase.master.TestMasterFailover in HBase-0.92-security >>> build 5) but not in insecure build. >>> >>> We also need to tackle the usual hanging build issue on Apache servers. >>> e.g. my attempt of triggering TRUNK builds 2469 and 2470 failed with: >>> >>> Caused by: java.lang.LinkageError: loader (instance of >>> hudson/remoting/RemoteClassLoader): attempted duplicate class definition >>> for name: "hudson/model/Descriptor" >>> at java.lang.ClassLoader.defineClass1(Native Method) >>> at java.lang.ClassLoader.defineClass(ClassLoader.java:634) >>> at java.lang.ClassLoader.defineClass(ClassLoader.java:480) >>> >>> If we consider the upcoming 0.94(-security) builds, the scale of >>> maintaining good build gets larger. >>> >>> Your wisdom of tackling this problem is welcome. >>> >> >> >