Adds support for the Kudu DECIMAL type introduced in Kudu 1.7.0.
Note: Adding support for Kudu decimal min/max filters is
tracked in IMPALA-6533.
Tests:
* Added Kudu create with decimal test to AnalyzeDDLTest.java
* Added Kudu table_format to test_decimal_queries.py
** Both decimal.test and decimal-exprs.test workloads
* Added decimal queries to the following Kudu workloads:
** kudu_create.test
** kudu_delete.test
** kudu_insert.test
** kudu_update.test
** kudu_upsert.test
Change-Id: I3a9fe5acadc53ec198585d765a8cfb0abe56e199
Reviewed-on: http://gerrit.cloudera.org:8080/9368
Reviewed-by: Dimitris Tsirogiannis <dtsirogiannis@cloudera.com>
Tested-by: Impala Public Jenkins
This change adds support for "clustered", "noclustered", "shuffle" and
"noshuffle" hints in CTAS statement.
Example:
create /*+ clustered,noshuffle */ table t partitioned by (year, month)
as select * from functional.alltypes
The effect of these hints are the same as in insert statements:
clustered:
Sort locally by partition expression before insert to ensure that only
one partition is written at a time. The goal is to reduce the number of
files kept open / buffers kept in memory simultaneously.
noclustered:
Do not sort by primary key before insert to Kudu table. No effect on HDFS
tables currently, as this is their default behavior.
shuffle:
Forces the planner to add an exchange node that repartitions by the
partition expression of the output table. This means that a partition
will be written only by a single node, which minimizes the global
number of simultaneous writes.
If only one partition is written (because all partitioning columns
are constant or the target table is not partitioned), then the shuffle
hint leads to a plan where all rows are merged at the coordinator where
the table sink is executed.
noshuffle:
Do not add exchange node before insert to partitioned tables.
The parser needed some modifications to be able to differentiate between
CREATE statements that allow hints (CTAS), and CREATE statements that
do not (every other type of CREATE statements). As a result, KW_CREATE
was moved from tbl_def_without_col_defs to statement rules.
Testing:
The parser tests mirror the tests of INSERT, while analysis and planner
tests are minimal, as the underlying logic is the same as for INSERT.
Query tests are not created, as the hints have no effect on
the DDL part of CTAS, and the actual query ran is the same as in
the insert case.
Change-Id: I8d74bca999da8ae1bb89427c70841f33e3c56ab0
Reviewed-on: http://gerrit.cloudera.org:8080/8400
Reviewed-by: Alex Behm <alex.behm@cloudera.com>
Tested-by: Impala Public Jenkins
If the first number in a row group written by Impala is NaN,
then Impala writes incorrect statistics in the metadata.
This will result in incorrect results when filtering the
data.
This commit fixes the read path when encountering NaNs in
Parquet min/max statistics. If min and max are both NaN, we
can't use the statistics at all. If only one of them is NaN,
the other still can be used.
I added some tests to QueryTest/parqet-stats.test
Change-Id: If3897fc1426541239223670812f59e2bed32f455
Reviewed-on: http://gerrit.cloudera.org:8080/9358
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
The default value for rpc_max_message_size is 50MB.
Impala currently requires support for messages of
up to 2GB. This changes the value of rpc_max_message_size
to INT_MAX for Impala.
Testing:
- Added a test to test_very_large_strings that generates
a row with multiple large strings. This row requires
that the RPC framework successfully transmit over
400MB. This works for both KRPC and Thrift.
This query operates under the same amount of memory
as other queries in large_strings.test.
- Tested separately that larger row sizes also work,
including tests up to almost 2GB.
Change-Id: I876bba0536e1d85e41eacd9c0aeccfe5c2126e58
Reviewed-on: http://gerrit.cloudera.org:8080/9337
Reviewed-by: Joe McDonnell <joemcdonnell@cloudera.com>
Tested-by: Impala Public Jenkins
Previously, tuple pointers of a row batch are allocated from
the heap via malloc() and tuple data is allocated from the
MemPool associated with the RowBatch. This change converts
the allocations of tuple pointers and tuple data to using
BufferPool for row batches allocated from KrpcDataStreamRecvr.
The primary motivation for this change is to take advantage of
the fact that buffers allocated from BufferPool always go back
to the per-core arena they came from when they are freed. This
alleviates the TCMalloc imbalance between the RPC service threads
and the fragment execution threads. As described in IMPALA-5518,
row batches are always allocated from the service threads' TCMalloc
cache and placed into the fragment execution threads' TCMalloc cache
when they're freed. This leads to underflow and overflow in those
threads' caches and high contention for the spinlock of the central
free list. With BufferPool, the memory always went back to its
originating arena so this kind of imbalance is less likely to occur.
This also dovetails with the long term plan to put most allocations
under BufferPool and have each operators in the plan reserved
appropriate amount of memory before execution.
Note that the proper reservation mechanism of the exchange node
hasn't yet been implemented in this change so the buffer pool client
handle used for allocating buffers has an ad-hoc set-up of no reservation
limit and using root reservation tracker as parent. This needs to be
fixed as part of IMPALA-6524. The default buffer pool limit is also
bumped to 85% to account for the extra usage from the exchange nodes.
The minimum buffer size is also lowered to 8KB to reduce amount of memory
wastage as a row batch's tuple pointers / tuple data can sometimes be
much smaller than 64KB.
Testing done: Debug core build.
Change-Id: If4b1a45f68b9df0d3b539511e15aff15700246f2
Reviewed-on: http://gerrit.cloudera.org:8080/9344
Reviewed-by: Michael Ho <kwho@cloudera.com>
Tested-by: Impala Public Jenkins
One of the spilling test was failing because its minimum bufferpool
mem requirement was more when ran on local FS as compared to when
it is run on HDFS.
The fix is to increase the bufferpool limit to a value just above
the min limit so that it still forces spill to disk on both filesystems.
Testing:
Ran core tests with local FS as target file system. Made sure the
failing test passed.
Change-Id: I50648d7936007a26891cf64d6343c47d9d646596
Reviewed-on: http://gerrit.cloudera.org:8080/9354
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
Tuple pointers in the generated row batches may not be initialized
if a tuple has byte size 0. There are some codes which compare these
uninitialized pointers against nullptr so having them uninitialized
may return wrong (and non-deterministic) results, e.g.:
impala::TupleIsNullPredicate::GetBooleanVal()
The following query produces non-deterministic results currently:
SELECT count(v.x) FROM functional.alltypestiny t3 LEFT OUTER JOIN (
SELECT true AS x FROM functional.alltypestiny t1 LEFT OUTER JOIN
functional.alltypestiny t2 ON (true)) v ON (v.x = t3.bool_col)
WHERE t3.bool_col = true;
The alltypestiny table has 8 records, 4 records of them has the true
boolean value for bool_col. Therefore, the above query is a fancy
way of saying "8 * 8 * 4", i.e. it should return 256.
The solution is that scanners initialize tuple pointers to a non-null
value when there are no materialized slots. This non-null value is
provided by the new static member Tuple::POISON.
I extended QueryTest/scanners.test with the above query. This test
executes the query against all table formats.
This change has the biggest performance impact on count(*) queries on
large kudu tables. For my quick benchmark I copied tpch_kudu.lineitem
and doubled its data. The resulting table has 12,002,430 rows.
Without this patch 'select count(*) from biglineitem' runs for ~0.12s.
With the patch applied, the overhead is around a dozens of ms. I measured
the query on my desktop PC using a relase build of Impala. On debug builds,
the execution time of the patched version is around 160% of the original
version.
Without this patch:
+--------------+--------+----------+----------+--------+------------+-----------+---------------+---------------------+
| Operator | #Hosts | Avg Time | Max Time | #Rows | Est. #Rows | Peak Mem | Est. Peak Mem | Detail |
+--------------+--------+----------+----------+--------+------------+-----------+---------------+---------------------+
| 03:AGGREGATE | 1 | 127.50us | 127.50us | 1 | 1 | 28.00 KB | 10.00 MB | FINALIZE |
| 02:EXCHANGE | 1 | 22.32ms | 22.32ms | 3 | 1 | 0 B | 0 B | UNPARTITIONED |
| 01:AGGREGATE | 3 | 1.78ms | 1.89ms | 3 | 1 | 16.00 KB | 10.00 MB | |
| 00:SCAN KUDU | 3 | 8.00ms | 8.28ms | 12.00M | -1 | 512.00 KB | 0 B | default.biglineitem |
+--------------+--------+----------+----------+--------+------------+-----------+---------------+---------------------+
With this patch:
+--------------+--------+----------+----------+--------+------------+-----------+---------------+---------------------+
| Operator | #Hosts | Avg Time | Max Time | #Rows | Est. #Rows | Peak Mem | Est. Peak Mem | Detail |
+--------------+--------+----------+----------+--------+------------+-----------+---------------+---------------------+
| 03:AGGREGATE | 1 | 129.01us | 129.01us | 1 | 1 | 28.00 KB | 10.00 MB | FINALIZE |
| 02:EXCHANGE | 1 | 33.00ms | 33.00ms | 3 | 1 | 0 B | 0 B | UNPARTITIONED |
| 01:AGGREGATE | 3 | 1.99ms | 2.13ms | 3 | 1 | 16.00 KB | 10.00 MB | |
| 00:SCAN KUDU | 3 | 13.13ms | 13.97ms | 12.00M | -1 | 512.00 KB | 0 B | default.biglineitem |
+--------------+--------+----------+----------+--------+------------+-----------+---------------+---------------------+
Change-Id: I298122aaaa7e62eb5971508e0698e189519755de
Reviewed-on: http://gerrit.cloudera.org:8080/9239
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
The sender timed out error message diverges between Thrift and KRPC
slightly due to the source address being not readily available with
Thrift RPC implementation. This leads to failure in test_exchange_delays
when KRPC is enabled.
This change fixes the problem by shortening the error message string
to match against.
Testing done: Tested with KRPC enabled in the code and verified the tests passed.
Change-Id: Idd9410381dbb931231c92f084917265e5067b4c9
Reviewed-on: http://gerrit.cloudera.org:8080/9331
Reviewed-by: Sailesh Mukil <sailesh@cloudera.com>
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
There were a few places where we accidentally used fully-qualified
table references. As a result, the testTpchViews() test did not
exactly cover what was intended.
Change-Id: I886c451ab61a1739af96eeb765821dfd8e951b07
Reviewed-on: http://gerrit.cloudera.org:8080/9270
Reviewed-by: Taras Bobrovytsky <tbobrovytsky@cloudera.com>
Tested-by: Impala Public Jenkins
In EXPLAIN_LEVEL=2+, change the explain format for parquet predicate
statistics to output each tuple descriptor per line. This change is to
make it consistent with the output of other predicates.
Before:
parquet statistics predicates: c_custkey < 10, o_orderkey < 5, l_linenumber < 3
After:
parquet statistics predicates: c_custkey < 10
parquet statistics predicates on o: o_orderkey < 5
parquet statistics predicates on o_lineitems: l_linenumber < 3
Testing:
- Ran existing planner tests and updated the ones that are affected by
this change.
- Ran end-to-end tests in query_test
Change-Id: Ia3d55ab6a1ae551867a9f68b3622844102cc854e
Reviewed-on: http://gerrit.cloudera.org:8080/9223
Tested-by: Impala Public Jenkins
Reviewed-by: Alex Behm <alex.behm@cloudera.com>
This patch adds changes to the planner to account for memory used by
bloom filters at the fragment instance level. Also adds changes to
allocate memory for those bloom filters from the buffer pool.
Testing:
- Modified Planner Tests and end to end tests to account for memory
reservation for the runtime filters.
- Modified backend tests and benchmarks to use the bufferpool for
bloom filter allocation.
- Add an end to end test.
- Ran rest of the core tests.
Change-Id: Iea2759665fb2e8bef9433014a8d42a7ebf99ce1f
Reviewed-on: http://gerrit.cloudera.org:8080/8971
Reviewed-by: Bikramjeet Vig <bikramjeet.vig@cloudera.com>
Tested-by: Impala Public Jenkins
The encoding was added in an early version of the Parquet
spec and deprecated even in the Parquet 1.0 spec.
Parquet-MR switched to generating RLE at the same time as
the spec changed in mid-2013. Impala always wrote RLE:
see commit 6e293090e6.
The Impala implementation of BIT_PACKED was never correct
because it implemented little endian bit unpacking instead of
the big endian unpacking required by the spec for levels.
Testing:
Updated tests to reflect expected behaviour for supported
and unsupported def level encodings.
Cherry-picks: not for 2.x.
Change-Id: I12c75b7f162dd7de8e26cf31be142b692e3624ae
Reviewed-on: http://gerrit.cloudera.org:8080/9241
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
Previously, an analytic function that used the same expr in both the
'partition by' and 'order by' clauses, and where the expr meets the
criteria for being materialized before the sort, would hit an
IllegalStateException due to the expr being inserted into the same
ExprSubstitutionMap twice.
If the values have already been partitioned on the expr, then all of
the values for it in each partition will be the same and also ordering
on the expr doesn't change the results. So, the fix is to simply
exclude the duplicate expr from the 'order by' while still
partitioning on it.
Testing:
- Added a regression test to PlannerTest.
Change-Id: Id5f1d5fbc6f69df5850f96afed345ce27668c30b
Reviewed-on: http://gerrit.cloudera.org:8080/9218
Reviewed-by: Alex Behm <alex.behm@cloudera.com>
Tested-by: Impala Public Jenkins
Introduces a new TBLPROPERTY for controlling stats
extrapolation on a per-table basis:
impala.enable.stats.extrapolation=true/false
The property key was chosen to be consistent with
the impalad startup flag --enable_stats_extrapolation
and to indicate that the property was set and is used
by Impala.
Behavior:
- If the property is not set, then the extrapolation
behavior is determined by the impalad startup flag.
- If the property is set, it overrides the impalad
startup flag, i.e., extrapolation can be explicitly
enabled or disabled regardless of the startup flag.
Testing:
- added new unit tests
- code/hdfs run passed
Change-Id: Ie49597bf1b93b7572106abc620d91f199cba0cfd
Reviewed-on: http://gerrit.cloudera.org:8080/9139
Reviewed-by: Alex Behm <alex.behm@cloudera.com>
Tested-by: Impala Public Jenkins
This change enables clustering by default. IMPALA-2521 introduced the
'clustered' hint which inserts a local sort by the partitioning columns
to a query plan. The hint is only effective for HDFS and Kudu tables.
Like before, the 'noclustered' hint prevents clustering. If a table has
ordering columns defined, the 'noclustered' hint is ignored and we
issue a warning.
This change removes some tests that were added specifically to test
that clustering can be enabled using the 'clustered' hint. It changes
some tests to use the 'noclustered' hint to make sure that clustering
can be disabled. It also adds tests to make sure that we cover the
'noclustered' case properly.
Cherry-picks: not for 2.x.
Change-Id: Idbf2368cf4415e6ecfa65058daf6ff87ef62f9d9
Reviewed-on: http://gerrit.cloudera.org:8080/9153
Reviewed-by: Lars Volker <lv@cloudera.com>
Tested-by: Impala Public Jenkins
Change the expected result type of Kudu TPCH Q17 to Decimal because
DECIMAL_V2 is now enabled by default. This was not done earlier because
we were not running TPCH on Kudu regularly.
Cherry-picks: not for 2.x
Change-Id: I46fc038d40969547622707ce77a037494f0ed0a9
Reviewed-on: http://gerrit.cloudera.org:8080/9208
Reviewed-by: Taras Bobrovytsky <tbobrovytsky@cloudera.com>
Tested-by: Impala Public Jenkins
Based on the existing Parquet column chunk level statistics null_count,
Impala's Parquet scanner is enhanced to skip an entire row group if the
null_count statistics indicate that all the values under the predicated
column are NULL as we wouldn't get any result rows from that row group
anyway.
Change-Id: I141317af0e0df30da8f220b29b0bfba364f40ddf
Reviewed-on: http://gerrit.cloudera.org:8080/9140
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
Currently the catalog data is compressed in the statestore, but
uncompressed when passed between FE and BE. It results in a ~2GB limit
on the metadata. IMPALA-3499 introduced a workaround in the impalad but
there isn't one in the catalogd. This patch aims to increase the size
limit for statestore updates, reduce the copying of the metadata and
reduce the memory footprint. With this patch, the catalog objects are
passed and (de)compressed between FE and BE one at a time. The new
limits are:
- A single catalog object cannot be larger than ~2GB.
- A statestore catalog update cannot be larger than ~4GB. It is
compressed size if FLAGS_compact_catalog_topic is true.
The behavior of the catalog op executer is not changed. The data is not
compressed and the size limit is still 2GB.
Testing: Ran existing tests. A test for compressing and decompressing
catalog objects is added. Manually tested with a 1.95GB catalog object
and a 3.90 GB uncompressed statestore update.
Change-Id: I3a8819cad734b3a416eef6c954e55b73cc6023ae
Reviewed-on: http://gerrit.cloudera.org:8080/8825
Reviewed-by: Tianyi Wang <twang@cloudera.com>
Tested-by: Impala Public Jenkins
This patch reserves SQL:2016 reserved words, excluding:
1. Impala builtin function names.
2. Time unit words(year, month, etc.).
3. An exception list based on a discussion.
Some test cases are modified to avoid these words. A impalad and
catalogd startup option reserved_words_version is added. The words are
reserved if the option is set to "3.0.0".
Change-Id: If1b295e6a77e840cf1b794c2eb73e1b9d2b8ddd6
Reviewed-on: http://gerrit.cloudera.org:8080/9096
Reviewed-by: Alex Behm <alex.behm@cloudera.com>
Reviewed-by: Philip Zeyliger <philip@cloudera.com>
Tested-by: Impala Public Jenkins
The 'compute stats' statement currently computes column-level
statistics for all columns of a table.
This adds potentially unneeded work for columns whose stats
are not needed by queries. It can be especially costly for
very wide tables and unneeded large string fields.
This change modifies the 'compute stats' (non-incremental only)
to support a user-specified list of columns for which stats
should be computed. An example with the extension is as follows:
compute stats my_db.my_table(column_a, column_b);
While the phrase "for columns ..." is commonly used, since
'compute stats' seems fairly unique (vs. 'analyze table ...'),
this change favors brevity with the parenthesized column list.
Whereas currently 'compute stats' is applied to the columns that
can be analyzed, the 'compute stats' in this change results in
an error when a column is specified that cannot be analyzed
(e.g., column does not exist, column is of an unsupported type,
column is a partitioning column). Moreover, an empty column
list can be supplied which means that no columns will be analyzed.
Testing:
- analyzing a subset of columns is already supported (e.g., not all
columns can be analyzed), so the focus with testing is to check
that the user-specified columns are handled as expected.
- tests include: parser tests, ddl analysis, end-to-end tests.
Change-Id: If8b25dd248e578dc7ddd35468125cca12d1b9f27
Reviewed-on: http://gerrit.cloudera.org:8080/9133
Reviewed-by: Alex Behm <alex.behm@cloudera.com>
Tested-by: Impala Public Jenkins
Adds a concept of a "removed" query option that has no effect but does
not return an error when a user attempts to set it. These options are
not returned by "set" or "set all" commands that are executed in
impala-shell or server-side.
These query options have been deprecated for several releases:
DEFAULT_ORDER_BY_LIMIT, ABORT_ON_DEFAULT_LIMIT_EXCEEDED,
V_CPU_CORES, RESERVATION_REQUEST_TIMEOUT, RM_INITIAL_MEM,
SCAN_NODE_CODEGEN_THRESHOLD, MAX_IO_BUFFERS
RM_INITIAL_MEM did still have an effect, but it was undocumented and
MEM_LIMIT should be used in preference.
DISABLE_CACHED_READS also had an effect but it was documented as
deprecated.
Otherwise the options had no effect at all.
Testing:
Ran exhaustive build.
Updated query option tests to reflect the new behaviour.
Cherry-picks: not for 2.x.
Change-Id: I9e742e9b0eca0e5c81fd71db3122fef31522fcad
Reviewed-on: http://gerrit.cloudera.org:8080/9118
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
Escapes the following special characters in RE2 library:
.\+*?[^]$(){}=!<>|:-
Testing:
Add some unit tests into ExprTest.StringRegexpFunctions
Add some E2E tests into exprs.test
Change-Id: I84c3e0ded26f6eb20794c38b75be9b25cd111e4b
Reviewed-on: http://gerrit.cloudera.org:8080/8900
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
In this commit we enable Decimal_V2 by default. We also update the
expected results in many of our tests.
Testing:
Ran an exhaustive test which almost passed. Updated the few failed
tests in it.
Cherry-pick: not for 2.x
Change-Id: Ibbdd05bf986b7947f106b396017faa3a0bd87fd7
Reviewed-on: http://gerrit.cloudera.org:8080/9062
Reviewed-by: Taras Bobrovytsky <tbobrovytsky@cloudera.com>
Tested-by: Impala Public Jenkins
We should not perform alias substitution in the
subexpressions of GROUP BY, HAVING, and ORDER BY
to be more standard conformant.
=== Allowed ===
SELECT int_col / 2 AS x
FROM functional.alltypes
GROUP BY x;
SELECT int_col / 2 AS x
FROM functional.alltypes
ORDER BY x;
SELECT NOT bool_col AS nb
FROM functional.alltypes
GROUP BY nb
HAVING nb;
=== Not allowed ===
SELECT int_col / 2 AS x
FROM functional.alltypes
GROUP BY x / 2;
SELECT int_col / 2 AS x
FROM functional.alltypes
ORDER BY -x;
SELECT int_col / 2 AS x
FROM functional.alltypes
GROUP BY x
HAVING x > 3;
Some extra checks were added to AnalyzeExprsTest.java.
I had to update other tests to make them pass
since the new behavior is more restrictive.
I added alias.test to the end-to-end tests.
Cherry-picks: not for 2.x.
Change-Id: I0f82483b486acf6953876cfa672b0d034f3709a8
Reviewed-on: http://gerrit.cloudera.org:8080/8801
Reviewed-by: Alex Behm <alex.behm@cloudera.com>
Tested-by: Impala Public Jenkins
Currently we do not codegen CHAR types. This change checks
for CHAR literals in a expr and disables codegen.
Change-Id: I7e4e27350c53bc69ce412a004e392e7480214f73
Reviewed-on: http://gerrit.cloudera.org:8080/9102
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
This change disallows explicitly setting the Kudu table name property
for managed Kudu tables in a CREATE TABLE statement. The Kudu table
name property gets a generated value as the following:
'impala::db_name.table_name' where table_name is the one given in
the CREATE TABLE statement.
Providing the Kudu table name property when creating a managed Kudu
table results in an error without creating the table. E.g.:
CREATE TABLE t (i INT) STORED AS KUDU
TBLPROPERTIES('kudu.table_name'='some_name');
Alongside the CREATE TABLE statement also the ALTER TABLE statement
is changed not to allow the modification of Kudu.table_name of
managed Kudu tables.
Change-Id: Ieca037498abf8f5fde67b77e824b720482cdbe6f
Reviewed-on: http://gerrit.cloudera.org:8080/8820
Reviewed-by: Dimitris Tsirogiannis <dtsirogiannis@cloudera.com>
Tested-by: Impala Public Jenkins
Currently, top-level scalar columns in parquet files can
be used at runtime to prune row-groups by evaluating certain
conjuncts over the column's dictionary (if available).
This change extends such pruning to scalar values that are
stored in collection type columns. Currently, dictionary
pruning works by finding eligible conjuncts for top-level
slots. Since only top-level slots are supported, the slots
are implicitly part of the scan node's tuple descriptor.
With this change, we track eligible conjuncts by slot as well
as the tuple that contains the slot (either top-level or
nested collection). Since collection conjuncts are already
managed by a map that associates tuple descriptors to a list
of their conjuncts, this extension follows the existing
representation.
The frontend builds the mapping of SlotId to conjuncts that
are dictionary filterable. This mapping now includes SlotId's
that reference nested tuples. The backend is adjusted to
use the same representation. In addition, collection
readers are decomposed into scalar filterable columns and
other, non-dictionary filterable readers. When filtering
a row group using a conjunct associated to a (possibly)
nested collection type, an additional tuple buffer is
allocated per tuple descriptor.
Testing:
- e2e test extended to illustrate row-groups that are pruned
by nested collection dictionary filters.
Change-Id: If3a2abcfc3d0f7d18756816659fed77ce12668dd
Reviewed-on: http://gerrit.cloudera.org:8080/8775
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
Previously it had to be executed serially because it modified tables in
the functional database.
This change separates out tests that use temporary tables and runs those
in a unique_database.
Testing:
Ran locally in a loop with parallelism of 4 for a while.
Change-Id: I2f62ede90f619b8cebbb1276bab903e7555d9744
Reviewed-on: http://gerrit.cloudera.org:8080/9022
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
Currently the default and min spillable buffer size and max row size
query options accept any valid int64 value. Since the planner depends
on these values for memory estimations, if a very large value close to
the limits of int64 is set, the variables representing or relying on
these estimates can overflow during different phases of query execution.
This patch puts a reasonable upper limit of 1TB to these query options
to prevent such a situation.
Testing:
Added backend query option tests.
Change-Id: I36d3915f7019b13c3eb06f08bfdb38c71ec864f1
Reviewed-on: http://gerrit.cloudera.org:8080/9023
Reviewed-by: Bikramjeet Vig <bikramjeet.vig@cloudera.com>
Tested-by: Impala Public Jenkins
We add new TPCDS .test files that are expected to be run with decimal_v2
enabled. The new expected results were generated using Impala and I
inspected them manually.
Change-Id: Ib867c51a521ec4a087bc127d99aee4b95ba97733
Reviewed-on: http://gerrit.cloudera.org:8080/8985
Reviewed-by: Michael Brown <mikeb@cloudera.com>
Tested-by: Impala Public Jenkins
Dataload currently executes bin/load-data.py for TPC-H,
TPC-DS, and functional-query concurrently. One of the final
steps for bin/load-data.py is to run a global "invalidate
metadata". Global "invalidate metadata" commands are known
to cause problem on concurrent systems. See IMPALA-5087.
For dataload, if TPC-H executes "invalidate metadata" while
TPC-DS is still creating tables and adding partitions,
the TPC-DS executor might erroneously believe that a table
does not exist.
This changes dataload to invalidate metadata at an
individual table level rather than globally. This
prevents the concurrency issue.
This also changes the names of some of the intermediate
SQL files generated by generate-schema-statements.py
and consumed by load-data.py to make them less confusing.
Change-Id: Ibc3a6d8a674a0bf6b02069bfe8a5e12034335b1f
Reviewed-on: http://gerrit.cloudera.org:8080/9009
Reviewed-by: Joe McDonnell <joemcdonnell@cloudera.com>
Tested-by: Impala Public Jenkins
SnappyDecompressor::MaxOutputLen assumes the input pointer to be
non-null. It's not true when the parquet file is corrupted and the
compressed_page_size field in a page header is 0. This patch handles
this error instead of failing a DCHECK.
Testing: A bad parquet file with 0 compressed_page_size is added. It
crashes impala without this patch.
Change-Id: I0d42937aab92a74f8e104d2f7fcd64dc24f6a500
Reviewed-on: http://gerrit.cloudera.org:8080/8977
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
Before this patch, we would estimate the number of hosts for the union
node by looking only at the first union operand. This is obviously
incorrect and lead us to underestimate the value.
We fix the problem by setting the estimate to be the maximum of its
children.
Testing:
- Added a planner test that reproduces the issue
Change-Id: I51e1ecca8dbc84b2b5a72708667b2799d00279f0
Reviewed-on: http://gerrit.cloudera.org:8080/9017
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
Examples of new command:
ALTER TABLE t1 SET ROW FORMAT DELIMITED FIELDS TERMINATED BY '\002';
ALTER TABLE t1 SET ROW FORMAT DELIMITED LINES TERMINATED BY '\001';
Testing:
Added parser tests and unit tests for alter statements including
partition options.
Change-Id: I96e347463504915a6f33932552e4d1f61e9b1154
Reviewed-on: http://gerrit.cloudera.org:8080/8928
Reviewed-by: Alex Behm <alex.behm@cloudera.com>
Tested-by: Impala Public Jenkins
Due to the way in which we instantiate fair scheduler allocation
loader, we donot read the config overrides from the HDFS config
files.
This is an unexpected behavior from users' POV since we typically
support overrides like custom user -> group mapping via HDFS
config (for ex: LDAPGroupsMapping) that eventually affects the
query -> pool assignment.
Fix: This patch loads the hadoop default configuration so that the
underlying QueuePlacementPolicy is based on user specified overrides.
Testing (manual): Changed the core-site.xml to use LDAPGroupsMapping
instead of the default ShellBasedUnixGroupsMapping and confirmed that
the correct group mapping plugin is loaded, by adding additional logging.
Also, modified TestRequestPoolService to assert that the core-site xml
overrides are loaded.
Change-Id: Ibb93870c0cc37e2432a643a274931f1d3d13fb96
Reviewed-on: http://gerrit.cloudera.org:8080/9000
Reviewed-by: Bharath Vissapragada <bharathv@cloudera.com>
Tested-by: Impala Public Jenkins
murmur_hash relys on HashUtil::MurmurHash2_64 which MurmurHash2 64-bit
version.
Testing:
Add unit tests for primitive types: ExprTest.MurmurHashFunction
Add E2E tests into exprs.test
Change-Id: I14d56ffb8fab256f3f66a2669271fd4b3c50cc29
Reviewed-on: http://gerrit.cloudera.org:8080/8893
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
When there are too many digits to the right of the dot in a decimal, we
would always truncate when casting to timestamp. In this patch we change
the behavior to round instead of truncating when decimal_v2 is enabled.
Testing:
- Added some EE tests, ran BE tests on my machine.
Change-Id: I8fb3a7d976ab980b8572d7e9524850572bad57da
Reviewed-on: http://gerrit.cloudera.org:8080/8969
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
Allow to specify Oracle-style hint on INSERT/UPSERT statements. For example,
- insert /* +noshuffle */ into table functional.alltypes partition(year,
month) select * from functional.alltypes;
- upsert /* +noshuffle */ into functional_kudu.alltypes select * from
functional.alltypes;
Testing:
Add unit tests to ParserTest#TestPlanHints
Add plan check tests to PlannerTest#testInsert, PlannerTest#testKuduUpsert
Add tests to ToSqlTest#planHintsTest
Change-Id: Ied7629d70197a0270cdc0853e00cc021fdb4dc20
Reviewed-on: http://gerrit.cloudera.org:8080/8676
Reviewed-by: Dimitris Tsirogiannis <dtsirogiannis@cloudera.com>
Tested-by: Impala Public Jenkins
This patch maps a signed integer logical type in parquet to a supported
Impala column type. This change introduces the following mapping -
INT_8 -> TINYINT
INT_16 -> SMALLINT
INT_32 -> INT
INT_64 -> BIGINT
Also, added a parquet file with the following schema for testing -
schema {
optional int32 id;
optional int32 tinyint_col (INT_8);
optional int32 smallint_col (INT_16);
optional int32 int_col;
optional int64 bigint_col;
}
Change-Id: I47a8371858c9597c6a440808cf6f933532468927
Reviewed-on: http://gerrit.cloudera.org:8080/8548
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Reviewed-by: Tianyi Wang <twang@cloudera.com>
Tested-by: Impala Public Jenkins
When the data loading finishes, it is possible for some HDFS blocks to
be under replicated. If impala gets the metadata before the replication
is done, some tests may fail. This patch adds a replication waiting step
in the data loading script.
Resubmitted with filesystem type check.
Change-Id: I64d9a8ea1d0a32b40047321b50a7139a8f48eac8
Reviewed-on: http://gerrit.cloudera.org:8080/8916
Reviewed-by: Vuk Ercegovac <vercegovac@cloudera.com>
Reviewed-by: Alex Behm <alex.behm@cloudera.com>
Tested-by: Impala Public Jenkins
When materialising a nested collection, has_template_tuple() should use
the template tuple for the collection, not the top-level tuple.
Testing:
Added tests based on nested-types-basic.test that operate on a simple
partitioned table. The tests reliably crashed Impala before the fix.
Change-Id: Ic808b824ce3b31af0539036d8ca23d17b18deab4
Reviewed-on: http://gerrit.cloudera.org:8080/8947
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
This patch fixes several issues related to the min/max aggregate
functions and their handling of 'nan' and 'inf':
- Previously, if 'inf' or '-inf' was the only value for the min/max
and codegen was being used, the result would be incorrect. This
occurred, for example in the case of 'inf' and 'min', because we
set an initial value of numeric_limits::max, which is less than
'inf', so the returned min was numeric_limits::max when it should be
'inf'. The fix is to set the initial value to
numeric_limits::infinity.
- Previously, if one of the values was 'nan', the result of min/max
was non-deterministic depending on the order the values were
evaluated in. This occurs because 'nan' < or > 'any value' is always
false, so if the first value added was 'nan', all other comparisons
would be false and 'nan' would be returned, whereas if the first
value wasn't 'nan' then the 'nan' wouldn't be returned. The fix is
to treat 'nan' specially and to always return 'nan' if there is a
single 'nan' value.
Testing:
- Added e2e tests for both scenarios, as well as adding a little extra
nan/inf coverage for other aggregate functions.
Change-Id: Ia1e206105937ce5afc75ca5044597d39b3dc6a81
Reviewed-on: http://gerrit.cloudera.org:8080/8854
Reviewed-by: Bikramjeet Vig <bikramjeet.vig@cloudera.com>
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
This patch adds a query rewriter to remove redundant explicit casts to
a string type (string, char, varchar) from binary predicates of the form
"cast(<non-const expr> to <string type>) <eq/ne op> <string constant>".
The cast is redundant if the predicate evaluation is the same even if
the cast is removed and the constant is converted to the original type
of the expression. For example:
cast(int_col as string) = '123456' -> int_col = 123456
Performance:
For the following query on a table having 6001215 records -
select * from tpch.lineitem where cast(l_linenumber as string) = '0'
+-----------------+-----------+--------+
| | Scan Time |
+-----------------+-----------+--------+
| | Avg | St dev |
| Without rewrite | 1s406ms | 44ms |
| With rewrite | 1s099ms | 28ms |
+-----------------+-----------+--------+
Testing:
- Added unit tests to ExprRewriteRulesTest
- Added functional test to expr.test
- Current FE planner tests and BE expr-test run successfully with this
change.
Change-Id: I91b7c6452d0693115f9b9ed9ba09f3ffe0f36b2b
Reviewed-on: http://gerrit.cloudera.org:8080/8660
Reviewed-by: Dimitris Tsirogiannis <dtsirogiannis@cloudera.com>
Tested-by: Impala Public Jenkins
As a follow-on to centralizing into one parent pom, we can now manage
thirdparty dependency versions in Java a little bit more clearly.
Upgrades SLF4J, commons.io:
slf4j: 1.7.5 -> 1.7.25
commons.io: 2.4 -> 2.6
The SLF4J upgrade is nice to be able to run under Java9. The release
notes at https://www.slf4j.org/news.html are uneventful.
Commons IO 2.6 supports Java 9 and is source and binary compatible,
per https://commons.apache.org/proper/commons-io/upgradeto2_6.html and
https://commons.apache.org/proper/commons-io/upgradeto2_5.html.
Removes the following dependencies:
htrace-core
hadoop-mapreduce-client-core
hive-shims
com.stumbleupon:async
commons-dbcp
jdo-api
I ran "mvn dependency:analyze" and these were some (but not all)
of the "Unused declared dependencies found." Spelunking in git logs,
these dependencies are from 2013 and possibly from an effort
to run with dependencies from the filesystem. They don't seem
to be required anymore.
Stops pulling in an old version of hadoop-client and kite-data-core in
testdata/TableFlattener by using the same versions as the Hadoop we use.
Doing so was unnecessarily causing us to download extra, old Hadoop
jars, and the new Hadoop jars seem to work just as well. This is the
kind of divergence that centralizing the versions into variables will
help with.
Creates variables for:
junit.version
slf4j.version
hadoop.version
commons-io.version
httpcomponents.core.version
thrift.version
kite.version (controlled via $IMPALA_KITE_VERSION in impala-config.sh)
Cleans up unused IMPALA_PARQUET_URL variables in impala-config.sh. We
only download Parquet via Maven, rather than downloading it in the
toolchain, so this variable wasn't doing anything.
I ran the core tests with this change.
Change-Id: I717e0625dfe0fdbf7e9161312e9e80f405a359c5
Reviewed-on: http://gerrit.cloudera.org:8080/8853
Reviewed-by: Joe McDonnell <joemcdonnell@cloudera.com>
Tested-by: Impala Public Jenkins
Using fsck breaks non-HDFS builds: local, S3, and Isilon.
This reverts commit 5a7c10ec3d.
Change-Id: I0b12a42049543ca0b267b5146a0bbcdd2316abfc
Reviewed-on: http://gerrit.cloudera.org:8080/8880
Reviewed-by: Michael Brown <mikeb@cloudera.com>
Tested-by: Impala Public Jenkins
Modifies COMPUTE STATS TABLESAMPLE to use the new SAMPLED_NDV()
function.
Testing:
- modified/improved existing functional tests
- core/hdfs run passed
Change-Id: I6ec0831f77698695975e45ec0bc0364c765d819b
Reviewed-on: http://gerrit.cloudera.org:8080/8840
Reviewed-by: Alex Behm <alex.behm@cloudera.com>
Tested-by: Impala Public Jenkins
When the data loading finishes, it is possible for some HDFS blocks to
be under replicated. If impala gets the metadata before the replication
is done, some tests may fail. This patch adds a replication waiting step
in the data loading script.
Change-Id: I88dfb7165b7515b3e96111436be490f2068ec322
Reviewed-on: http://gerrit.cloudera.org:8080/8846
Reviewed-by: Alex Behm <alex.behm@cloudera.com>
Tested-by: Impala Public Jenkins
We saw some failures on the exhaustive release build because the
compiler assumed that the pointer to the intermediate struct that is
used for computing decimal average was aligned.
To fix the problem, we mark the struct with a "packed" attribute so
that the compiler does not expect it to be aligned.
Testing:
- Ran the failing test locally on an release build and it passed.
Change-Id: Id25ec6e20dde3f50fb37a22135b355ad251809e0
Reviewed-on: http://gerrit.cloudera.org:8080/8836
Reviewed-by: Taras Bobrovytsky <tbobrovytsky@cloudera.com>
Tested-by: Impala Public Jenkins
