[
https://issues.apache.org/jira/browse/HIVE-12724?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15098971#comment-15098971
]
Eugene Koifman commented on HIVE-12724:
---------------------------------------
made a Blocker since this is data loss
> ACID: Major compaction fails to include the original bucket files into MR job
> -----------------------------------------------------------------------------
>
> Key: HIVE-12724
> URL: https://issues.apache.org/jira/browse/HIVE-12724
> Project: Hive
> Issue Type: Bug
> Components: Hive, Transactions
> Affects Versions: 1.3.0, 2.0.0
> Reporter: Wei Zheng
> Assignee: Wei Zheng
> Priority: Blocker
> Attachments: HIVE-12724.1.patch, HIVE-12724.2.patch,
> HIVE-12724.3.patch, HIVE-12724.4.patch, HIVE-12724.ADDENDUM.1.patch,
> HIVE-12724.branch-1.2.patch, HIVE-12724.branch-1.patch
>
>
> How the problem happens:
> * Create a non-ACID table
> * Before non-ACID to ACID table conversion, we inserted row one
> * After non-ACID to ACID table conversion, we inserted row two
> * Both rows can be retrieved before MAJOR compaction
> * After MAJOR compaction, row one is lost
> {code}
> hive> USE acidtest;
> OK
> Time taken: 0.77 seconds
> hive> CREATE TABLE t1 (nationkey INT, name STRING, regionkey INT, comment
> STRING)
> > CLUSTERED BY (regionkey) INTO 2 BUCKETS
> > STORED AS ORC;
> OK
> Time taken: 0.179 seconds
> hive> DESC FORMATTED t1;
> OK
> # col_name data_type comment
> nationkey int
> name string
> regionkey int
> comment string
> # Detailed Table Information
> Database: acidtest
> Owner: wzheng
> CreateTime: Mon Dec 14 15:50:40 PST 2015
> LastAccessTime: UNKNOWN
> Retention: 0
> Location: file:/Users/wzheng/hivetmp/warehouse/acidtest.db/t1
> Table Type: MANAGED_TABLE
> Table Parameters:
> transient_lastDdlTime 1450137040
> # Storage Information
> SerDe Library: org.apache.hadoop.hive.ql.io.orc.OrcSerde
> InputFormat: org.apache.hadoop.hive.ql.io.orc.OrcInputFormat
> OutputFormat: org.apache.hadoop.hive.ql.io.orc.OrcOutputFormat
> Compressed: No
> Num Buckets: 2
> Bucket Columns: [regionkey]
> Sort Columns: []
> Storage Desc Params:
> serialization.format 1
> Time taken: 0.198 seconds, Fetched: 28 row(s)
> hive> dfs -ls /Users/wzheng/hivetmp/warehouse/acidtest.db;
> Found 1 items
> drwxr-xr-x - wzheng staff 68 2015-12-14 15:50
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1
> hive> dfs -ls /Users/wzheng/hivetmp/warehouse/acidtest.db/t1;
> hive> INSERT INTO TABLE t1 VALUES (1, 'USA', 1, 'united states');
> WARNING: Hive-on-MR is deprecated in Hive 2 and may not be available in the
> future versions. Consider using a different execution engine (i.e. tez,
> spark) or using Hive 1.X releases.
> Query ID = wzheng_20151214155028_630098c6-605f-4e7e-a797-6b49fb48360d
> Total jobs = 1
> Launching Job 1 out of 1
> Number of reduce tasks determined at compile time: 2
> In order to change the average load for a reducer (in bytes):
> set hive.exec.reducers.bytes.per.reducer=<number>
> In order to limit the maximum number of reducers:
> set hive.exec.reducers.max=<number>
> In order to set a constant number of reducers:
> set mapreduce.job.reduces=<number>
> Job running in-process (local Hadoop)
> 2015-12-14 15:51:58,070 Stage-1 map = 100%, reduce = 100%
> Ended Job = job_local73977356_0001
> Loading data to table acidtest.t1
> MapReduce Jobs Launched:
> Stage-Stage-1: HDFS Read: 0 HDFS Write: 0 SUCCESS
> Total MapReduce CPU Time Spent: 0 msec
> OK
> Time taken: 2.825 seconds
> hive> dfs -ls /Users/wzheng/hivetmp/warehouse/acidtest.db/t1;
> Found 2 items
> -rwxr-xr-x 1 wzheng staff 112 2015-12-14 15:51
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/000000_0
> -rwxr-xr-x 1 wzheng staff 472 2015-12-14 15:51
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/000001_0
> hive> SELECT * FROM t1;
> OK
> 1 USA 1 united states
> Time taken: 0.434 seconds, Fetched: 1 row(s)
> hive> ALTER TABLE t1 SET TBLPROPERTIES ('transactional' = 'true');
> OK
> Time taken: 0.071 seconds
> hive> DESC FORMATTED t1;
> OK
> # col_name data_type comment
> nationkey int
> name string
> regionkey int
> comment string
> # Detailed Table Information
> Database: acidtest
> Owner: wzheng
> CreateTime: Mon Dec 14 15:50:40 PST 2015
> LastAccessTime: UNKNOWN
> Retention: 0
> Location: file:/Users/wzheng/hivetmp/warehouse/acidtest.db/t1
> Table Type: MANAGED_TABLE
> Table Parameters:
> COLUMN_STATS_ACCURATE false
> last_modified_by wzheng
> last_modified_time 1450137141
> numFiles 2
> numRows -1
> rawDataSize -1
> totalSize 584
> transactional true
> transient_lastDdlTime 1450137141
> # Storage Information
> SerDe Library: org.apache.hadoop.hive.ql.io.orc.OrcSerde
> InputFormat: org.apache.hadoop.hive.ql.io.orc.OrcInputFormat
> OutputFormat: org.apache.hadoop.hive.ql.io.orc.OrcOutputFormat
> Compressed: No
> Num Buckets: 2
> Bucket Columns: [regionkey]
> Sort Columns: []
> Storage Desc Params:
> serialization.format 1
> Time taken: 0.049 seconds, Fetched: 36 row(s)
> hive> set hive.support.concurrency=true;
> hive> set hive.txn.manager=org.apache.hadoop.hive.ql.lockmgr.DbTxnManager;
> hive> set hive.compactor.initiator.on=true;
> hive> set hive.compactor.worker.threads=5;
> hive> set hive.exec.dynamic.partition.mode=nonstrict;
> hive> dfs -ls /Users/wzheng/hivetmp/warehouse/acidtest.db/t1;
> Found 2 items
> -rwxr-xr-x 1 wzheng staff 112 2015-12-14 15:51
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/000000_0
> -rwxr-xr-x 1 wzheng staff 472 2015-12-14 15:51
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/000001_0
> hive> INSERT INTO TABLE t1 VALUES (2, 'Canada', 1, 'maple leaf');
> WARNING: Hive-on-MR is deprecated in Hive 2 and may not be available in the
> future versions. Consider using a different execution engine (i.e. tez,
> spark) or using Hive 1.X releases.
> Query ID = wzheng_20151214155028_630098c6-605f-4e7e-a797-6b49fb48360d
> Total jobs = 1
> Launching Job 1 out of 1
> Number of reduce tasks determined at compile time: 2
> In order to change the average load for a reducer (in bytes):
> set hive.exec.reducers.bytes.per.reducer=<number>
> In order to limit the maximum number of reducers:
> set hive.exec.reducers.max=<number>
> In order to set a constant number of reducers:
> set mapreduce.job.reduces=<number>
> Job running in-process (local Hadoop)
> 2015-12-14 15:54:18,943 Stage-1 map = 100%, reduce = 100%
> Ended Job = job_local1674014367_0002
> Loading data to table acidtest.t1
> MapReduce Jobs Launched:
> Stage-Stage-1: HDFS Read: 0 HDFS Write: 0 SUCCESS
> Total MapReduce CPU Time Spent: 0 msec
> OK
> Time taken: 1.995 seconds
> hive> dfs -ls /Users/wzheng/hivetmp/warehouse/acidtest.db/t1;
> Found 3 items
> -rwxr-xr-x 1 wzheng staff 112 2015-12-14 15:51
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/000000_0
> -rwxr-xr-x 1 wzheng staff 472 2015-12-14 15:51
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/000001_0
> drwxr-xr-x - wzheng staff 204 2015-12-14 15:54
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/delta_0000007_0000007_0000
> hive> dfs -ls
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/delta_0000007_0000007_0000;
> Found 2 items
> -rw-r--r-- 1 wzheng staff 214 2015-12-14 15:54
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/delta_0000007_0000007_0000/bucket_00000
> -rw-r--r-- 1 wzheng staff 797 2015-12-14 15:54
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/delta_0000007_0000007_0000/bucket_00001
> hive> SELECT * FROM t1;
> OK
> 1 USA 1 united states
> 2 Canada 1 maple leaf
> Time taken: 0.1 seconds, Fetched: 2 row(s)
> hive> ALTER TABLE t1 COMPACT 'MAJOR';
> Compaction enqueued.
> OK
> Time taken: 0.026 seconds
> hive> show compactions;
> OK
> Database Table Partition Type State Worker Start Time
> Time taken: 0.022 seconds, Fetched: 1 row(s)
> hive> dfs -ls /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/;
> Found 3 items
> -rwxr-xr-x 1 wzheng staff 112 2015-12-14 15:51
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/000000_0
> -rwxr-xr-x 1 wzheng staff 472 2015-12-14 15:51
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/000001_0
> drwxr-xr-x - wzheng staff 204 2015-12-14 15:55
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/base_0000007
> hive> dfs -ls /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/base_0000007;
> Found 2 items
> -rw-r--r-- 1 wzheng staff 222 2015-12-14 15:55
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/base_0000007/bucket_00000
> -rw-r--r-- 1 wzheng staff 802 2015-12-14 15:55
> /Users/wzheng/hivetmp/warehouse/acidtest.db/t1/base_0000007/bucket_00001
> hive> select * from t1;
> OK
> 2 Canada 1 maple leaf
> Time taken: 0.396 seconds, Fetched: 1 row(s)
> hive> select count(*) from t1;
> WARNING: Hive-on-MR is deprecated in Hive 2 and may not be available in the
> future versions. Consider using a different execution engine (i.e. tez,
> spark) or using Hive 1.X releases.
> Query ID = wzheng_20151214155028_630098c6-605f-4e7e-a797-6b49fb48360d
> Total jobs = 1
> Launching Job 1 out of 1
> Number of reduce tasks determined at compile time: 1
> In order to change the average load for a reducer (in bytes):
> set hive.exec.reducers.bytes.per.reducer=<number>
> In order to limit the maximum number of reducers:
> set hive.exec.reducers.max=<number>
> In order to set a constant number of reducers:
> set mapreduce.job.reduces=<number>
> Job running in-process (local Hadoop)
> 2015-12-14 15:56:20,277 Stage-1 map = 100%, reduce = 100%
> Ended Job = job_local1720993786_0003
> MapReduce Jobs Launched:
> Stage-Stage-1: HDFS Read: 0 HDFS Write: 0 SUCCESS
> Total MapReduce CPU Time Spent: 0 msec
> OK
> 1
> Time taken: 1.623 seconds, Fetched: 1 row(s)
> {code}
> Note, the cleanup doesn't kick in because the compaction fails already. The
> cleanup itself doesn't have any problem (at least not that we know of for
> this case).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
