[
https://issues.apache.org/jira/browse/CALCITE-3202?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16895261#comment-16895261
]
Kai Guo commented on CALCITE-3202:
----------------------------------
I just put the same test on Mongo adaptor, it is surprise that the test passed.
Compared the code of ElasticsearchAggregate and MongoAggregate, the difference
is that MongoAggregate constructor doesn't contain code which
ElasticsearchAggregate constructor already had as below
{code:java}
if (getConvention() != input.getConvention()) {
String message = String.format(Locale.ROOT, "%s != %s", getConvention(),
input.getConvention());
throw new AssertionError(message);
}
{code}
And after deleting these code in ElasticsearchAggregate, test passed. It is
obvious that the easiest way to solve this problem is just delete the code, but
since most of the ElasticSearch and Mongo adaptor's rel constructor contains
code like
{code:java}
assert getConvention() == child.getConvention();
{code}
when vm parameter _-ea_ added, the test may still fail.
Thinking as an adaptor, it is okey to make sure that the convention of input
should be the same with itself. As optimizer, it will ensure the rel finally
get the right convention, so it seems that adaptor doesn't need to worry about
the input's convention that much.
Since other adaptors may also have problem like this, I think the 1 way is
better.
> AssertionError in ElasticsearchAggregate constructor when applying
> AggregateProjectMergeRule
> --------------------------------------------------------------------------------------------
>
> Key: CALCITE-3202
> URL: https://issues.apache.org/jira/browse/CALCITE-3202
> Project: Calcite
> Issue Type: Bug
> Components: elasticsearch-adapter
> Affects Versions: 1.21.0
> Reporter: Kai Guo
> Priority: Critical
> Attachments: log.log.zip, mapping.json
>
>
> Thanks for viewing this issue! The error occurs when I was trying to execute
> a query with some subquery:
> {code:java}
> select distinct uni_id
> from (
> select uni_id from sample_index where customer_child = 'customer' and
> customer_from_plat = 'FOO'
> and uni_id in (select uni_id from sample_index where customer_child =
> 'trade' and shop_id = '60790435')
> and uni_id in (select uni_id from sample_index where customer_child =
> 'member' and member_id = '884225534')
> )
> {code}
> While running code below:
> {code:java}
> ElasticsearchAggregate(RelOptCluster cluster,
> RelTraitSet traitSet,
> RelNode input,
> ImmutableBitSet groupSet,
> List<ImmutableBitSet> groupSets,
> List<AggregateCall> aggCalls) throws InvalidRelException {
> super(cluster, traitSet, input, groupSet, groupSets, aggCalls);
> if (getConvention() != input.getConvention()) {
> String message = String.format(Locale.ROOT, "%s != %s", getConvention(),
> input.getConvention());
> throw new AssertionError(message);
> }{code}
> an AssertionError which means an input of ElasticsearchAggregate has a
> Convention of NONE throws out. stack traces shows as below:
> {code:java}
> java.lang.AssertionError: ELASTICSEARCH != NONE
> at
> org.apache.calcite.adapter.elasticsearch.ElasticsearchAggregate.<init>(ElasticsearchAggregate.java:66)
> at
> org.apache.calcite.adapter.elasticsearch.ElasticsearchAggregate.copy(ElasticsearchAggregate.java:112)
> at
> org.apache.calcite.rel.rules.AggregateProjectMergeRule.apply(AggregateProjectMergeRule.java:113)
> at
> org.apache.calcite.rel.rules.AggregateProjectMergeRule.onMatch(AggregateProjectMergeRule.java:72)
> at
> org.apache.calcite.plan.volcano.VolcanoRuleCall.onMatch(VolcanoRuleCall.java:208)
> at
> org.apache.calcite.plan.volcano.VolcanoPlanner.findBestExp(VolcanoPlanner.java:631)
> at org.apache.calcite.tools.Programs.lambda$standard$3(Programs.java:283)
> at org.apache.calcite.tools.Programs$SequenceProgram.run(Programs.java:343)
> at org.apache.calcite.prepare.Prepare.optimize(Prepare.java:189)
> at org.apache.calcite.prepare.Prepare.prepareSql(Prepare.java:320)
> at org.apache.calcite.prepare.Prepare.prepareSql(Prepare.java:231)
> at
> org.apache.calcite.prepare.CalcitePrepareImpl.prepare2_(CalcitePrepareImpl.java:637)
> at
> org.apache.calcite.prepare.CalcitePrepareImpl.prepare_(CalcitePrepareImpl.java:501)
> at
> org.apache.calcite.prepare.CalcitePrepareImpl.prepareSql(CalcitePrepareImpl.java:471)
> at
> org.apache.calcite.jdbc.CalciteConnectionImpl.parseQuery(CalciteConnectionImpl.java:231)
> at
> org.apache.calcite.jdbc.CalciteConnectionImpl.prepareStatement_(CalciteConnectionImpl.java:213)
> at
> org.apache.calcite.jdbc.CalciteConnectionImpl.prepareStatement(CalciteConnectionImpl.java:202)
> at
> org.apache.calcite.jdbc.CalciteConnectionImpl.prepareStatement(CalciteConnectionImpl.java:93)
> at
> org.apache.calcite.avatica.AvaticaConnection.prepareStatement(AvaticaConnection.java:175)
> at SearchTest.main(SearchTest.java:25)
> {code}
> And before the volcano planner start to work, the algebra expression looks
> like this:
> {code:java}
> LogicalAggregate(group=[{0}])
> LogicalProject(uni_id=[$2])
> LogicalFilter(condition=[AND(=($0, 'customer'), =($1, 'FOO'))])
> LogicalJoin(condition=[=($2, $4)], joinType=[inner])
> LogicalJoin(condition=[=($2, $3)], joinType=[inner])
> LogicalProject(customer_child=[CAST(ITEM($0,
> 'customer_child')):VARCHAR(65535)], customer_from_plat=[CAST(ITEM($0,
> 'customer_from_plat')):VARCHAR(255)], uni_id=[CAST(ITEM($0,
> 'uni_id')):VARCHAR(255)])
> ElasticsearchTableScan(table=[[elasticsearch_raw, sample_index]])
> LogicalAggregate(group=[{0}])
> LogicalProject(uni_id=[$2])
> LogicalFilter(condition=[AND(=($0, 'trade'), =($1,
> '60790435'))])
> LogicalProject(customer_child=[CAST(ITEM($0,
> 'customer_child')):VARCHAR(65535)], shop_id=[CAST(ITEM($0,
> 'shop_id')):VARCHAR(255)], uni_id=[CAST(ITEM($0, 'uni_id')):VARCHAR(255)])
> ElasticsearchTableScan(table=[[elasticsearch_raw,
> sample_index]])
> LogicalAggregate(group=[{0}])
> LogicalProject(uni_id=[$2])
> LogicalFilter(condition=[AND(=($0, 'member'), =($1,
> '884225534'))])
> LogicalProject(customer_child=[CAST(ITEM($0,
> 'customer_child')):VARCHAR(65535)], member_id=[CAST(ITEM($0,
> 'member_id')):VARCHAR(255)], uni_id=[CAST(ITEM($0, 'uni_id')):VARCHAR(255)])
> ElasticsearchTableScan(table=[[elasticsearch_raw,
> sample_index]]){code}
> Attachment:
> log.log.zip is a compressed log file with DEBUG level.
> mapping.json is the ES sample_index's mapping information.
> Please check them if needed.
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
