bug: QueryFailed: [1104]index out of bounds: the len is 16 but the index is 16 [v0.33.7]

[rad-pat]

Search before asking

• I had searched in the issues and found no similar issues.

Version

1.2.891

What's Wrong?

INSERT statement fails with QueryFailed: [1104]index out of bounds: the len is 16 but the index is 16 [v0.33.7]

How to Reproduce?

SQL is as below. We had recently upgraded from 1.2.790 to 1.2.891 and this error started appearing

INSERT INTO "s1"."target" (
	"RespCenterName"
	,"LineItemActivityName"
	,"ACC&SC Combo"
	,"ActivityDriver1Name"
	,"CostObject1Name"
	,"CostObject2Name"
	,"Include"
	,"PriorMonth"
	)
SELECT result."RespCenterName"
	,result."LineItemActivityName"
	,result."ACC&SC Combo"
	,result."ActivityDriver1Name"
	,result."CostObject1Name"
	,result."CostObject2Name"
	,result."Include"
	,result."PriorMonth"
FROM (
	SELECT CAST("PriorMonth_Combine"."RespCenterName" AS VARCHAR) AS "RespCenterName"
		,CAST("PriorMonth_Combine"."LineItemActivityName" AS VARCHAR) AS "LineItemActivityName"
		,CAST("PriorMonth_Combine"."ACC&SC Combo" AS VARCHAR) AS "ACC&SC Combo"
		,CAST(coalesce(anon_1."ActivityDriverName", "PriorMonth_Combine"."ActivityDriver1Name") AS VARCHAR) AS "ActivityDriver1Name"
		,CAST(coalesce(anon_1."CostObject1", "PriorMonth_Combine"."CostObject1Name") AS VARCHAR) AS "CostObject1Name"
		,CAST(coalesce(anon_1."CostObject2", "PriorMonth_Combine"."CostObject2Name") AS VARCHAR) AS "CostObject2Name"
		,CAST(coalesce(anon_1."Include", '1') AS VARCHAR) AS "Include"
		,CAST(coalesce(anon_1."PriorMonth", "PriorMonth_Combine"."PriorMonth") AS VARCHAR) AS "PriorMonth"
	FROM "s1"."source" AS "PriorMonth_Combine"
	LEFT OUTER JOIN (
		SELECT CAST("Weight"."RespCenterName" AS VARCHAR) AS "RespCenterName"
			,CAST("Weight"."ActivityName" AS VARCHAR) AS "ActivityName"
			,CAST(min("Weight"."ActivityDriverName") AS VARCHAR) AS "ActivityDriverName"
			,CAST(min("Weight"."CostObject1") AS VARCHAR) AS "CostObject1"
			,CAST(min("Weight"."CostObject2") AS VARCHAR) AS "CostObject2"
			,CAST(min("Weight"."Include") AS VARCHAR) AS "Include"
			,CAST(min("Weight"."FileName") AS VARCHAR) AS "FileName"
			,CAST(min("Weight"."PriorMonth") AS VARCHAR) AS "PriorMonth"
		FROM "s1"."source2" AS "Weight"
		WHERE EXISTS (
				SELECT *
				FROM "s1"."source" AS a_exists
				WHERE "Weight"."RespCenterName" = a_exists."RespCenterName"
					AND "Weight"."ActivityName" = a_exists."LineItemActivityName"
				)
		GROUP BY "Weight"."RespCenterName"
			,"Weight"."ActivityName"
		) AS anon_1 ON "PriorMonth_Combine"."RespCenterName" = anon_1."RespCenterName"
		AND "PriorMonth_Combine"."LineItemActivityName" = anon_1."ActivityName"
	GROUP BY "PriorMonth_Combine"."RespCenterName"
		,"PriorMonth_Combine"."LineItemActivityName"
		,"PriorMonth_Combine"."ACC&SC Combo"
		,CAST(coalesce(anon_1."ActivityDriverName", "PriorMonth_Combine"."ActivityDriver1Name") AS VARCHAR)
		,CAST(coalesce(anon_1."CostObject1", "PriorMonth_Combine"."CostObject1Name") AS VARCHAR)
		,CAST(coalesce(anon_1."CostObject2", "PriorMonth_Combine"."CostObject2Name") AS VARCHAR)
		,CAST(coalesce(anon_1."Include", '1') AS VARCHAR)
		,CAST(coalesce(anon_1."PriorMonth", "PriorMonth_Combine"."PriorMonth") AS VARCHAR)
	) AS result
WHERE result."PriorMonth" = 'Yes'

Are you willing to submit PR?

• Yes I am willing to submit a PR!

[KKould]

Thanks for reporting this.

I tried to reproduce this locally with the same query shape on v1.2.891-nightly, including:

• enable_experimental_new_join = 1
• max_block_size = 16
• synthetic data covering matched/unmatched rows, NULL values, LEFT JOIN, correlated EXISTS, GROUP BY, and COALESCE

However, I could not reproduce the index out of bounds: the len is 16 but the index is 16 error with synthetic data. The query completed successfully in my local tests.

Could you please provide a minimal reproducible case with desensitized DDL and sample data for the involved tables?

Ideally, please include:

• CREATE TABLE statements for s1.source, s1.source2, and s1.target
• A small set of desensitized INSERT statements or data files that can reproduce the issue
• Any non-default settings used when running the query

The error looks like an internal execution/block indexing issue rather than a table index issue, but we need a reproducible dataset to identify the exact operator path.

[rad-pat]

Non default settings are:

enable_distributed_compact	1	0	LOCAL	Enables distributed execution of table compaction.
enable_distributed_replace_into	1	0	LOCAL	Enables distributed execution of 'REPLACE INTO'.
enable_last_snapshot_location_hint	0	1	LOCAL	Enables writing last_snapshot_location_hint object
http_handler_result_timeout_secs	120	60	LOCAL	Set the timeout in seconds that a http query session expires without any polls.
max_execute_time_in_seconds	7200	0	GLOBAL	Sets the maximum query execution time in seconds. Setting it to 0 means no limit.
max_memory_usage	118111600640	202682698956	LOCAL	Sets the maximum memory usage in bytes for processing a single query.
max_storage_io_requests	10	64	LOCAL	Sets the maximum number of concurrent storage I/O requests.

setting for max_block_size was default == 65536

I will try to extract the data as requested, but this may take some time. I had already moved them to mainstream release 1.2.881 to see if it made any difference, but they may be wound back to 1.2.790 as it was running prior to the upgrade (not sure if that is helpful)