----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/36302/#review90915 -----------------------------------------------------------
Ship it! Thanks for clarifying some of the issues I raised. I think that if we can add documentation to the areas in which I was unclear about, then this patch looks good to me. - Jonathan Hurley On July 8, 2015, 11:10 a.m., Dmitro Lisnichenko wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/36302/ > ----------------------------------------------------------- > > (Updated July 8, 2015, 11:10 a.m.) > > > Review request for Ambari, Andrew Onischuk, Jonathan Hurley, and Nate Cole. > > > Bugs: AMBARI-12224 > https://issues.apache.org/jira/browse/AMBARI-12224 > > > Repository: ambari > > > Description > ------- > > HBase has deprecated hbase.regionserver.global.memstore.upperLimit and > hbase.regionserver.global.memstore.lowerLimit for HDP-2.3 in favor of > hbase.regionserver.global.memstore.size and > hbase.regionserver.global.memstore.size.lower.limit. > The slider called "% of RegionServer Allocated to Write Buffers" sets > hbase.regionserver.global.memstore.size, while > hbase.regionserver.global.memstore.upperLimit is in the advanced > configuration section. This leaves two bugs: > In HDP-2.2, the slider has no affect since HBase basically ignores the > parameter. > In HDP-2.3 the value set by the slider also has no affect since the > deprecated values set take precedence. > One solution is to get rid of hbase.regionserver.global.memstore.upperLimit > and hbase.regionserver.global.memstore.lowerLimit in advanced settings in > Ambari, and make the slider configuration to use > hbase.regionserver.global.memstore.upperLimit in HDP-2.2 and > hbase.regionserver.global.memstore.size in HDP-2.3. Rolling upgrade can > convert these settings. > > > Diffs > ----- > > > ambari-server/src/main/java/org/apache/ambari/server/serveraction/upgrades/HBaseConfigCalculation.java > PRE-CREATION > > ambari-server/src/main/resources/stacks/HDP/2.2/services/HBASE/themes/theme.json > 53da86d > ambari-server/src/main/resources/stacks/HDP/2.2/upgrades/upgrade-2.3.xml > c5fb156 > > ambari-server/src/main/resources/stacks/HDP/2.3/services/HBASE/configuration/hbase-site.xml > 12cd44b > ambari-server/src/main/resources/stacks/HDP/2.3/services/HBASE/metainfo.xml > 921a225 > > ambari-server/src/main/resources/stacks/HDP/2.3/services/HBASE/themes/theme.json > PRE-CREATION > > ambari-server/src/main/resources/stacks/HDPWIN/2.2/services/HBASE/themes/theme.json > 1808809 > > ambari-server/src/main/resources/stacks/HDPWIN/2.3/services/HBASE/configuration/hbase-site.xml > 43f872a > > ambari-server/src/main/resources/stacks/HDPWIN/2.3/services/HBASE/metainfo.xml > aa169df > > ambari-server/src/main/resources/stacks/HDPWIN/2.3/services/HBASE/themes/theme.json > PRE-CREATION > > Diff: https://reviews.apache.org/r/36302/diff/ > > > Testing > ------- > > manual checks > Tests in error: > UpgradeResourceProviderTest.testCreatePartialDowngrade:428 ? NullPointer > > Tests run: 3111, Failures: 0, Errors: 1, Skipped: 28 > > [INFO] > ------------------------------------------------------------------------ > [INFO] Reactor Summary: > [INFO] > [INFO] Ambari Views ...................................... SUCCESS [6.409s] > [INFO] Ambari Metrics Common ............................. SUCCESS [3.909s] > [INFO] Ambari Server ..................................... FAILURE > [1:16:42.534s] > [INFO] Ambari Agent ...................................... SKIPPED > [INFO] > ------------------------------------------------------------------------ > [INFO] BUILD FAILURE > [INFO] > ------------------------------------------------------------------------ > [INFO] Total time: 1:16:57.150s > [INFO] Finished at: Wed Jul 08 10:34:40 UTC 2015 > [INFO] Final Memory: 39M/826M > [INFO] > ------------------------------------------------------------------------ > > > Thanks, > > Dmitro Lisnichenko > >
