Files
Nikita Fomichev a78bcc708e utils: add auto-bisect script for ClickHouse regression bisection
Adds `utils/auto-bisect/` — a shell-based bisection framework that
downloads pre-built CI binaries and runs a user-supplied test script
to find the first bad commit.

Key features:
- Downloads public CI binaries from S3 by default; private CloudFront
  builds with `--private` (credentials via `CH_CI_BASIC_AUTH` env var,
  `CH_CI_USER`/`CH_CI_PASSWORD`, or `/etc/netrc`)
- Three environment modes: `single` (one server), `replicateddb` (two
  nodes + Keeper, Replicated database), `sharedcatalog` (two nodes +
  Keeper, SharedCatalog)
- Uses `git bisect --no-checkout` by default so the work tree is not
  modified during bisection
- Walker mode (`--walker`) for linear commit-by-commit testing
- Example test scripts in `tests/`

Tested locally: `--env nothing`, `--env single`, and `--env replicateddb`
all verified working.
2026-03-28 10:39:03 +01:00

46 lines
651 B
Bash
Executable File

#!/bin/bash
set -e
echo "$PWD"
CH_PATH=${CH_PATH:=clickhouse}
$CH_PATH client -q "SELECT version()";
$CH_PATH client -mn -q "
drop table if exists t1;
drop table if exists t2;
CREATE TABLE t1
(
c1 UInt64,
c2 UInt64,
c3 Nullable(UInt64)
)
engine = MergeTree()
PRIMARY KEY (c1, c2)
ORDER BY (c1, c2);
CREATE TABLE t2
(
c1 UInt64,
c2 UInt64,
c3 UInt64,
)
engine = MergeTree()
PRIMARY KEY (c1, c2)
ORDER BY (c1, c2);
insert into t1 select * from generateRandom() limit 10000;
insert into t2 select * from generateRandom() limit 10000;
SELECT
count()
FROM t1
INNER JOIN t2 ON t1.c3 = t2.c3
WHERE
has([], t1.c3)
"