> On Nov. 2, 2017, 10:40 p.m., Peter Vary wrote: > > standalone-metastore/src/main/java/org/apache/hadoop/hive/metastore/HiveAlterHandler.java > > Lines 265 (patched) > > <https://reviews.apache.org/r/63528/diff/1/?file=1879637#file1879637line265> > > > > Shouldn't we use > > parts = msdb.getPartitions(dbname, name, partitionBatchSize); > > instead > > parts = msdb.getPartitions(dbname, name, -1); > > > > This way we will load only the required amount of partitions, instead > > of all. > > Adam Szita wrote: > Although msdb.getPartitions(dbname, name, partitionBatchSize) only loads > partitionBatchSize amount of partitions, it will always return the same set > of partitions. This is used in dropPartitions and it's not a problem there as > we're also removing the requested set, so it is guaranteed to get new ones in > the next call. For our use case here it is sadly not feasable to use this > call. Taking a look at RawStore class I can see no iterative way to query the > partitions from the DB behind HMS. > > Peter Vary wrote: > Still think it would be useful, to have the possibility to load only > partial partition list into memory to decrease memory load. Shall we open a > follow-up jira for this? > Thanks, > Peter
Looks like a fine idea, I've opened HIVE-17987 for this one. - Adam ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/63528/#review189979 ----------------------------------------------------------- On Nov. 2, 2017, 10:02 p.m., Adam Szita wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/63528/ > ----------------------------------------------------------- > > (Updated Nov. 2, 2017, 10:02 p.m.) > > > Review request for hive, Peter Vary and Barna Zsombor Klara. > > > Bugs: HIVE-17969 > https://issues.apache.org/jira/browse/HIVE-17969 > > > Repository: hive-git > > > Description > ------- > > Refactoring alter table code to use batching of partitions when calling the > heavy removeUnusedColumnDescriptor method > > > Diffs > ----- > > > standalone-metastore/src/main/java/org/apache/hadoop/hive/metastore/HiveAlterHandler.java > ccadac1ada6aaae884ab39f5d99e91b8c542404e > > standalone-metastore/src/main/java/org/apache/hadoop/hive/metastore/ObjectStore.java > 62801c53853dbafb7c425cff943ec819dcee4800 > > > Diff: https://reviews.apache.org/r/63528/diff/1/ > > > Testing > ------- > > > Thanks, > > Adam Szita > >
