Split test into 2

This commit is contained in:
Mikhail Filimonov 2020-10-22 11:02:14 +02:00
parent 111b553ee5
commit 77f66e5a09
No known key found for this signature in database
GPG Key ID: 6E49C2E9AF1220BE
4 changed files with 104 additions and 48 deletions

View File

@ -1,8 +0,0 @@
1 2020-01-01 String
2 2020-02-02 Another string
3 2020-03-03 One more string
4 2020-01-02 String for first partition
1 2020-01-01 String
2 2020-02-02 Another string
3 2020-03-03 One more string
4 2020-01-02 String for first partition

View File

@ -4,48 +4,10 @@ CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
. "$CURDIR"/../shell_config.sh . "$CURDIR"/../shell_config.sh
WORKING_FOLDER_01527="${CLICKHOUSE_TMP}/01527_clickhouse_local_optimize" WORKING_FOLDER_01527="${CLICKHOUSE_TMP}/01527_clickhouse_local_optimize"
rm -rf "${WORKING_FOLDER_01527}" rm -rf "${WORKING_FOLDER_01527}"
mkdir -p "${WORKING_FOLDER_01527}/metadata/local/" mkdir -p "${WORKING_FOLDER_01527}"
# OPTIMIZE was crashing due to lack of temporary volume in local # OPTIMIZE was crashing due to lack of temporary volume in local
${CLICKHOUSE_LOCAL} --query "drop database if exists d; create database d; create table d.t engine MergeTree order by a as select 1 a; optimize table d.t final" -- --path="${WORKING_FOLDER_01527}" ${CLICKHOUSE_LOCAL} --query "drop database if exists d; create database d; create table d.t engine MergeTree order by a as select 1 a; optimize table d.t final" -- --path="${WORKING_FOLDER_01527}"
# Some extra (unrealted) scenarios of clickhouse-local usage.
## 1. Imagine we want to process this file:
cat <<EOF > "${WORKING_FOLDER_01527}/data.csv"
1,2020-01-01,"String"
2,2020-02-02,"Another string"
3,2020-03-03,"One more string"
4,2020-01-02,"String for first partition"
EOF
## 2. that is the metadata for the table we want to fill
## schema should match the schema of the table from server
## (the easiest way is just to copy it from the server)
cat <<EOF > "${WORKING_FOLDER_01527}/metadata/local/test.sql"
ATTACH TABLE local.test (id UInt64, d Date, s String) Engine=MergeTree ORDER BY id PARTITION BY toYYYYMM(d);
EOF
## 3a. that is the metadata for the input file we want to read
## it should match the structure of source file
## use stdin to read from pipe
cat <<EOF > "${WORKING_FOLDER_01527}/metadata/local/stdin.sql"
ATTACH TABLE local.stdin (id UInt64, d Date, s String) Engine=File(CSV, stdin);
EOF
## 3b. Instead of stdin you can use file path
cat <<EOF > "${WORKING_FOLDER_01527}/metadata/local/data_csv.sql"
ATTACH TABLE local.data_csv (id UInt64, d Date, s String) Engine=File(CSV, '${WORKING_FOLDER_01527}/data.csv');
EOF
## All preparations done, the rest is simple:
# option a (if 3a used) with pipe / reading stdin (truncate was added for the test)
cat "${WORKING_FOLDER_01527}/data.csv" | ${CLICKHOUSE_LOCAL} --query "INSERT INTO local.test SELECT * FROM local.stdin; SELECT * FROM local.test ORDER BY id; TRUNCATE TABLE local.test;" -- --path="${WORKING_FOLDER_01527}"
# option b (if 3b used) 0 with filepath (truncate was added for the test)
${CLICKHOUSE_LOCAL} --query "INSERT INTO local.test SELECT * FROM local.data_csv; SELECT * FROM local.test ORDER BY id; TRUNCATE TABLE local.test;" -- --path="${WORKING_FOLDER_01527}"
rm -rf "${WORKING_FOLDER_01527}" rm -rf "${WORKING_FOLDER_01527}"

View File

@ -0,0 +1,19 @@
Option 1. Prepare parts from from table with Engine=File defined in metadata, read from an arbitrary path
1 2020-01-01 String
2 2020-02-02 Another string
3 2020-03-03 One more string
4 2020-01-02 String for first partition
Option 2. Prepare parts from from table with Engine=File defined in metadata, read from stdin (pipe)
11 2020-01-01 String
12 2020-02-02 Another string
13 2020-03-03 One more string
14 2020-01-02 String for first partition
Option 3. Prepare parts from from table with Engine=File defined via command line, read from stdin (pipe)
21 2020-01-01 String
22 2020-02-02 Another string
23 2020-03-03 One more string
24 2020-01-02 String for first partition
Possibility to run optimize on prepared parts before sending parts to server
202001 1
202002 1
202003 1

View File

@ -0,0 +1,83 @@
#!/usr/bin/env bash
CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
. "$CURDIR"/../shell_config.sh
WORKING_FOLDER_01528="${CLICKHOUSE_TMP}/01527_clickhouse_local_optimize"
rm -rf "${WORKING_FOLDER_01528}"
mkdir -p "${WORKING_FOLDER_01528}/metadata/local"
## Checks scenario of preparing parts offline by clickhouse-local
## that is the metadata for the table we want to fill
## schema should match the schema of the table from server
## (the easiest way is just to copy it from the server)
cat <<EOF > "${WORKING_FOLDER_01528}/metadata/local/test.sql"
ATTACH TABLE local.test (id UInt64, d Date, s String) Engine=MergeTree ORDER BY id PARTITION BY toYYYYMM(d);
EOF
#################
echo "Option 1. Prepare parts from from table with Engine=File defined in metadata, read from an arbitrary path"
## Source file:
cat <<EOF > "${WORKING_FOLDER_01528}/data.csv"
1,2020-01-01,"String"
2,2020-02-02,"Another string"
3,2020-03-03,"One more string"
4,2020-01-02,"String for first partition"
EOF
## metadata written into file
cat <<EOF > "${WORKING_FOLDER_01528}/metadata/local/data_csv.sql"
ATTACH TABLE local.data_csv (id UInt64, d Date, s String) Engine=File(CSV, '${WORKING_FOLDER_01528}/data.csv');
EOF
## feed the table
${CLICKHOUSE_LOCAL} --query "INSERT INTO local.test SELECT * FROM local.data_csv;" -- --path="${WORKING_FOLDER_01528}"
## check the parts were created
${CLICKHOUSE_LOCAL} --query "SELECT * FROM local.test WHERE id < 10 ORDER BY id;" -- --path="${WORKING_FOLDER_01528}"
#################
echo "Option 2. Prepare parts from from table with Engine=File defined in metadata, read from stdin (pipe)"
cat <<EOF > "${WORKING_FOLDER_01528}/metadata/local/stdin.sql"
ATTACH TABLE local.stdin (id UInt64, d Date, s String) Engine=File(CSV, stdin);
EOF
cat <<EOF | ${CLICKHOUSE_LOCAL} --query "INSERT INTO local.test SELECT * FROM local.stdin;" -- --path="${WORKING_FOLDER_01528}"
11,2020-01-01,"String"
12,2020-02-02,"Another string"
13,2020-03-03,"One more string"
14,2020-01-02,"String for first partition"
EOF
${CLICKHOUSE_LOCAL} --query "SELECT * FROM local.test WHERE id BETWEEN 10 AND 19 ORDER BY id;" -- --path="${WORKING_FOLDER_01528}"
#################
echo "Option 3. Prepare parts from from table with Engine=File defined via command line, read from stdin (pipe)"
cat <<EOF | ${CLICKHOUSE_LOCAL} --query "INSERT INTO local.test SELECT * FROM table;" -S "id UInt64, d Date, s String" --input-format=CSV -- --path="${WORKING_FOLDER_01528}"
21,2020-01-01,"String"
22,2020-02-02,"Another string"
23,2020-03-03,"One more string"
24,2020-01-02,"String for first partition"
EOF
${CLICKHOUSE_LOCAL} --query "SELECT * FROM local.test WHERE id BETWEEN 20 AND 29 ORDER BY id;" -- --path="${WORKING_FOLDER_01528}"
#################
echo "Possibility to run optimize on prepared parts before sending parts to server"
${CLICKHOUSE_LOCAL} --query "OPTIMIZE TABLE local.test FINAL;" -- --path="${WORKING_FOLDER_01528}"
# ensure we have one part per partition
${CLICKHOUSE_LOCAL} --query "SELECT toYYYYMM(d) m, uniqExact(_part) FROM local.test GROUP BY m ORDER BY m" -- --path="${WORKING_FOLDER_01528}"
# cleanup
rm -rf "${WORKING_FOLDER_01528}"