[YSQL][ASH] Nested statements should be tracked only if pg_stat_statements tracks them #21031
Closed
1 task done
Assignees
Labels
Comments
abhinab-yb
added
area/ysql
yugabyte-ci
added
kind/bug
55 tasks
abhinab-yb
added a commit
that referenced
this issue
May 22, 2024
…ements is tracking it Summary: ASH's query id is the same as the one present in the pg_stat_statements view. In case of nested queries, pg_stat_statements may or may not track them, depending on the guc `pg_stat_statements.track`. This diff tracks the nested query ids with ASH only if pg_stat_statements is also doing it. This diff also avoids setting the query id in the post_parse_analyze hook, because in some cases of nested queries, only the post_parse_analyze hook is called and no subsequent ExecutorEnd / ProcessUtility hook is called, in which case we would push the query id to the stack but never pop it. Now, we start tracking the query id at the start of ExecutorStart / ProcessUtility hook and stop tracking it at the end of ExecutorEnd / ProcessUtility hook or if there is some error encountered in between. Jira: DB-10002, DB-11281 Test Plan: ./yb_build.sh --java-test TestYbAsh#testNestedQueriesWithAsh Reviewers: jason Reviewed By: jason Subscribers: hbhanawat, yql, amitanand Differential Revision: https://phorge.dev.yugabyte.com/D34773
svarnau
pushed a commit
that referenced
this issue
May 25, 2024
…ements is tracking it Summary: ASH's query id is the same as the one present in the pg_stat_statements view. In case of nested queries, pg_stat_statements may or may not track them, depending on the guc `pg_stat_statements.track`. This diff tracks the nested query ids with ASH only if pg_stat_statements is also doing it. This diff also avoids setting the query id in the post_parse_analyze hook, because in some cases of nested queries, only the post_parse_analyze hook is called and no subsequent ExecutorEnd / ProcessUtility hook is called, in which case we would push the query id to the stack but never pop it. Now, we start tracking the query id at the start of ExecutorStart / ProcessUtility hook and stop tracking it at the end of ExecutorEnd / ProcessUtility hook or if there is some error encountered in between. Jira: DB-10002, DB-11281 Test Plan: ./yb_build.sh --java-test TestYbAsh#testNestedQueriesWithAsh Reviewers: jason Reviewed By: jason Subscribers: hbhanawat, yql, amitanand Differential Revision: https://phorge.dev.yugabyte.com/D34773
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Jira Link: DB-10002
Description
ASH assigns a query id to each statement, it doesn't check if it's a nested statement or not. It can cause discrepancy when that same query id is not in pg_stat_statements.
If pg_stat_statements is only tracking the top level query, and not the nested statements, then the nested statement's samples might be ignored when joined based on query id.
Issue Type
kind/bug
Warning: Please confirm that this issue does not contain any sensitive information
The text was updated successfully, but these errors were encountered: