Can the SQLancer log of the github action run be obtained?

[hanyisong]

I want to find out why databend often gives the following error： https://github.com/hanyisong/sqlancer/actions/runs/3713158096/jobs/6295493367#step:6:307

Caused by: java.sql.SQLException: Code: 1003, displayText = Unknown database 'databend2'.

This type of error cannot be reproduced locally

[mrigger]

In this specific case, the log seems to be printed on stdout:

DROP DATABASE IF EXISTS databend2;
CREATE DATABASE databend2;
USE databend2;
CREATE TABLE t0(c0INT SMALLINT NOT NULL);
CREATE TABLE t1(c0VARCHAR VARCHAR NULL DEFAULT('B5lCyutb9'));
CREATE TABLE t2(c0INT TINYINT NULL);
CREATE TABLE t3(c0BOOLEAN BOOL NULL, c1VARCHAR VARCHAR NULL DEFAULT(''));
INSERT INTO t1(c0varchar) VALUES ('Ta');
INSERT INTO t1(c0varchar) VALUES ('1JklH'), ('');
INSERT INTO t3(c0boolean, c1varchar) VALUES (true, '5K');
INSERT INTO t3(c0boolean, c1varchar) VALUES (false, '9Jw');
INSERT INTO t2(c0int) VALUES (-1698699297), (820809467);
INSERT INTO t0(c0int) VALUES (-142044461), (-175630708), (1748266817);
INSERT INTO t3(c0boolean) VALUES (false);
INSERT INTO t1(c0varchar) VALUES ('lSXEj');
INSERT INTO t1(c0varchar) VALUES ('jWrP');
INSERT INTO t1(c0varchar) VALUES ('0pJ2NE');
INSERT INTO t1(c0varchar) VALUES ('8klbA22'), ('pfPYsa50');
INSERT INTO t1(c0varchar) VALUES ('');
INSERT INTO t2(c0int) VALUES (601162576);
INSERT INTO t3(c1varchar, c0boolean) VALUES ('TWbq', true), ('', false);
DELETE FROM t1 WHERE ((false)OR(true));
INSERT INTO t1(c0varchar) VALUES ('CJc9Y');
INSERT INTO t0(c0int) VALUES (618981459);
INSERT INTO t3(c1varchar, c0boolean) VALUES ('', false), ('', false), ('', true);
INSERT INTO t3(c1varchar, c0boolean) VALUES ('', false), ('4iCPnK', false), ('', true);
INSERT INTO t0(c0int) VALUES (214725767);
INSERT INTO t1(c0varchar) VALUES ('W5U');
INSERT INTO t1(c0varchar) VALUES ('jMyGTf');
INSERT INTO t1(c0varchar) VALUES ('Nm');
EXPLAIN SELECT t2.c0int, ((NULL) IS NOT NULL), true FROM t2 WHERE (((((NULL)/(t2.c0int)))+((- t2.c0int))) NOT IN (((((1409562065)+(t2.c0int)))/((- -1595066122))))) ORDER BY t2.c0int, t2.c0int, t2.c0int, t2.c0int;
INSERT INTO t2(c0int) VALUES (1698074387);
INSERT INTO t0(c0int) VALUES (585312142);
INSERT INTO t3(c1varchar, c0boolean) VALUES ('dsHLKBrD', false), ('W1', false);
INSERT INTO t3(c0boolean, c1varchar) VALUES (false, 'c1');
INSERT INTO t3(c0boolean, c1varchar) VALUES (false, 'KS');
SELECT SUM(count) FROM (SELECT ((false IS NOT NULL AND false) ::BIGINT)as count FROM t0) as res;

Not sure if that helps with reproducing the bug?

Bye the way, I noticed that there is no whitespace between the column name and data type (e.g., c0int). Is this intended?

[hanyisong]

This example cannot be reproduced, c0int is to prevent the join connection type from mismatching

[hanyisong]

The stack overflow error I found before was obtained from *-cur.log

[mrigger]

Not really sure how we could reproduce this. Perhaps you could open a PR that prints the relevant -cur.log file after the issue has been discovered (without intending to merge it), so you can obtain the relevant log file?