Merge pull request #79 from deep-stack/pm-single-node

Run migrations on a single-node TimescaleDB setup
This commit is contained in:
Ashwin Phatak 2022-05-17 16:19:34 +05:30 committed by GitHub
commit 14762a119f
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
14 changed files with 85 additions and 438 deletions

View File

@ -3,6 +3,14 @@ name: Docker Build
on: [pull_request]
jobs:
test:
# Add a dummy job to avoid failing GitHub CI checks.
# Other checks to be added later.
name: Dummy job
runs-on: ubuntu-latest
steps:
- name: Run dummy job
run: echo "Empty dummy job"
# concise_migration_diff:
# name: Verify concise migration and generated schema
# runs-on: ubuntu-latest

View File

@ -17,10 +17,8 @@ WORKDIR /app
COPY --from=builder /go/src/github.com/vulcanize/ipld-eth-db/scripts/startup_script.sh .
# copy over files for multi-node setup
COPY --from=builder /go/src/github.com/vulcanize/ipld-eth-db/docker-compose.test.yml docker-multi-node/docker-compose.test.yml
COPY --from=builder /go/src/github.com/vulcanize/ipld-eth-db/scripts/init-access-node.sh docker-multi-node/scripts/init-access-node.sh
COPY --from=builder /go/src/github.com/vulcanize/ipld-eth-db/scripts/init-data-node.sh docker-multi-node/scripts/init-data-node.sh
# copy over file for TimescaleDB setup
COPY --from=builder /go/src/github.com/vulcanize/ipld-eth-db/docker-compose.test.yml docker-tsdb/docker-compose.test.yml
COPY --from=builder /go/src/github.com/pressly/goose/cmd/goose/goose goose
COPY --from=builder /go/src/github.com/vulcanize/ipld-eth-db/db/migrations migrations/vulcanizedb

View File

@ -12,34 +12,28 @@ Schemas and utils for IPLD ETH Postgres database
docker-compose down -v --remove-orphans
```
* Spin up an access node and three data nodes using [docker-compose.test.yml](./docker-compose.test.yml):
* Spin up a TimescaleDB instance using [docker-compose.test.yml](./docker-compose.test.yml):
```bash
docker-compose -f docker-compose.test.yml up
```
Following final output should be seen on all the nodes:
Following final output should be seen:
```
LOG: TimescaleDB background worker launcher connected to shared catalogs
```
* Edit [startup_script.sh](./scripts/startup_script.sh) to change the number of migrations to be run:
```bash
./goose -dir migrations/vulcanizedb postgres "$VDB_PG_CONNECT" up-to 24
```
* In another `ipld-eth-db` terminal window, build an image `migrations-test` using [Dockerfile](./db/Dockerfile):
```bash
docker build -t migrations-test -f ./db/Dockerfile .
```
* After the access and data nodes have spun up, start a container using `migrations-test` image to run the db migrations:
* Start a container using `migrations-test` image to run the db migrations:
```bash
# Here, we are running the container using host network.
# So connect to access node on 127.0.0.1:8066
docker run --rm --network host -e DATABASE_USER=postgres -e DATABASE_PASSWORD=password -e DATABASE_HOSTNAME=127.0.0.1 -e DATABASE_PORT=8066 -e DATABASE_NAME=vulcanize_testing_v4 migrations-test
# So connect to TimescaleDB on 127.0.0.1:8066
docker run --rm --network host -e DATABASE_USER=vdbm -e DATABASE_PASSWORD=password -e DATABASE_HOSTNAME=127.0.0.1 -e DATABASE_PORT=8066 -e DATABASE_NAME=vulcanize_testing_v4 migrations-test
```

View File

@ -1,16 +0,0 @@
-- +goose NO TRANSACTION
-- +goose Up
-- this is generated by a script
SELECT add_data_node('dn3', host => 'data-node-3', port => 5432, database => 'vulcanize_testing_v4', password => 'password');
SELECT add_data_node('dn2', host => 'data-node-2', port => 5432, database => 'vulcanize_testing_v4', password => 'password');
SELECT add_data_node('dn1', host => 'data-node-1', port => 5432, database => 'vulcanize_testing_v4', password => 'password');
CALL distributed_exec($$ CREATE SCHEMA eth $$);
-- +goose Down
-- this is generated by a script
CALL distributed_exec($$ DROP SCHEMA eth $$);
SELECT delete_data_node('dn1', force => true, if_attached => true);
SELECT delete_data_node('dn2', force => true, if_attached => true);
SELECT delete_data_node('dn3', force => true, if_attached => true);

View File

@ -0,0 +1,68 @@
-- +goose Up
SELECT create_hypertable('public.blocks', 'block_number', migrate_data => true, chunk_time_interval => 32768);
SELECT create_hypertable('eth.header_cids', 'block_number', migrate_data => true, chunk_time_interval => 32768);
SELECT create_hypertable('eth.uncle_cids', 'block_number', migrate_data => true, chunk_time_interval => 32768);
SELECT create_hypertable('eth.transaction_cids', 'block_number', migrate_data => true, chunk_time_interval => 32768);
SELECT create_hypertable('eth.receipt_cids', 'block_number', migrate_data => true, chunk_time_interval => 32768);
SELECT create_hypertable('eth.state_cids', 'block_number', migrate_data => true, chunk_time_interval => 32768);
SELECT create_hypertable('eth.storage_cids', 'block_number', migrate_data => true, chunk_time_interval => 32768);
SELECT create_hypertable('eth.state_accounts', 'block_number', migrate_data => true, chunk_time_interval => 32768);
SELECT create_hypertable('eth.access_list_elements', 'block_number', migrate_data => true, chunk_time_interval => 32768);
SELECT create_hypertable('eth.log_cids', 'block_number', migrate_data => true, chunk_time_interval => 32768);
-- update version
INSERT INTO public.db_version (singleton, version) VALUES (true, 'v4.0.0-h')
ON CONFLICT (singleton) DO UPDATE SET (version, tstamp) = ('v4.0.0-h', NOW());
-- +goose Down
INSERT INTO public.db_version (singleton, version) VALUES (true, 'v4.0.0')
ON CONFLICT (singleton) DO UPDATE SET (version, tstamp) = ('v4.0.0', NOW());
-- reversing conversion to hypertable requires migrating all data from every chunk back to a single table
-- create new regular tables
CREATE TABLE eth.log_cids_i (LIKE eth.log_cids INCLUDING ALL);
CREATE TABLE eth.access_list_elements_i (LIKE eth.access_list_elements INCLUDING ALL);
CREATE TABLE eth.state_accounts_i (LIKE eth.state_accounts INCLUDING ALL);
CREATE TABLE eth.storage_cids_i (LIKE eth.storage_cids INCLUDING ALL);
CREATE TABLE eth.state_cids_i (LIKE eth.state_cids INCLUDING ALL);
CREATE TABLE eth.receipt_cids_i (LIKE eth.receipt_cids INCLUDING ALL);
CREATE TABLE eth.transaction_cids_i (LIKE eth.transaction_cids INCLUDING ALL);
CREATE TABLE eth.uncle_cids_i (LIKE eth.uncle_cids INCLUDING ALL);
CREATE TABLE eth.header_cids_i (LIKE eth.header_cids INCLUDING ALL);
CREATE TABLE public.blocks_i (LIKE public.blocks INCLUDING ALL);
-- migrate data
INSERT INTO eth.log_cids_i (SELECT * FROM eth.log_cids);
INSERT INTO eth.access_list_elements_i (SELECT * FROM eth.access_list_elements);
INSERT INTO eth.state_accounts_i (SELECT * FROM eth.state_accounts);
INSERT INTO eth.storage_cids_i (SELECT * FROM eth.storage_cids);
INSERT INTO eth.state_cids_i (SELECT * FROM eth.state_cids);
INSERT INTO eth.receipt_cids_i (SELECT * FROM eth.receipt_cids);
INSERT INTO eth.transaction_cids_i (SELECT * FROM eth.transaction_cids);
INSERT INTO eth.uncle_cids_i (SELECT * FROM eth.uncle_cids);
INSERT INTO eth.header_cids_i (SELECT * FROM eth.header_cids);
INSERT INTO public.blocks_i (SELECT * FROM public.blocks);
-- drop hypertables
DROP TABLE eth.log_cids;
DROP TABLE eth.access_list_elements;
DROP TABLE eth.state_accounts;
DROP TABLE eth.storage_cids;
DROP TABLE eth.state_cids;
DROP TABLE eth.receipt_cids;
DROP TABLE eth.transaction_cids;
DROP TABLE eth.uncle_cids;
DROP TABLE eth.header_cids;
DROP TABLE public.blocks;
-- rename new tables
ALTER TABLE eth.log_cids_i RENAME TO log_cids;
ALTER TABLE eth.access_list_elements_i RENAME TO access_list_elements;
ALTER TABLE eth.state_accounts_i RENAME TO state_accounts;
ALTER TABLE eth.storage_cids_i RENAME TO storage_cids;
ALTER TABLE eth.state_cids_i RENAME TO state_cids;
ALTER TABLE eth.receipt_cids_i RENAME TO receipt_cids;
ALTER TABLE eth.transaction_cids_i RENAME TO transaction_cids;
ALTER TABLE eth.uncle_cids_i RENAME TO uncle_cids;
ALTER TABLE eth.header_cids_i RENAME TO header_cids;
ALTER TABLE public.blocks_i RENAME TO blocks;

View File

@ -1,130 +0,0 @@
-- +goose Up
-- creating distributed hypertables from duplicate tables for now as we are getting the following error (while running geth
-- unit tests) if regular tables are directly converted to distributed hypertables
-- error: "cannot PREPARE a transaction that has executed LISTEN, UNLISTEN, or NOTIFY"
-- create new regular tables
CREATE TABLE eth.log_cids_i (LIKE eth.log_cids INCLUDING ALL);
CREATE TABLE eth.access_list_elements_i (LIKE eth.access_list_elements INCLUDING ALL);
CREATE TABLE eth.state_accounts_i (LIKE eth.state_accounts INCLUDING ALL);
CREATE TABLE eth.storage_cids_i (LIKE eth.storage_cids INCLUDING ALL);
CREATE TABLE eth.state_cids_i (LIKE eth.state_cids INCLUDING ALL);
CREATE TABLE eth.receipt_cids_i (LIKE eth.receipt_cids INCLUDING ALL);
CREATE TABLE eth.transaction_cids_i (LIKE eth.transaction_cids INCLUDING ALL);
CREATE TABLE eth.uncle_cids_i (LIKE eth.uncle_cids INCLUDING ALL);
CREATE TABLE eth.header_cids_i (LIKE eth.header_cids INCLUDING ALL);
CREATE TABLE public.blocks_i (LIKE public.blocks INCLUDING ALL);
-- turn them into distributed hypertables
SELECT create_distributed_hypertable('public.blocks_i', 'block_number', chunk_time_interval => 32768);
SELECT create_distributed_hypertable('eth.header_cids_i', 'block_number', chunk_time_interval => 32768);
SELECT create_distributed_hypertable('eth.uncle_cids_i', 'block_number', chunk_time_interval => 32768);
SELECT create_distributed_hypertable('eth.transaction_cids_i', 'block_number', chunk_time_interval => 32768);
SELECT create_distributed_hypertable('eth.receipt_cids_i', 'block_number', chunk_time_interval => 32768);
SELECT create_distributed_hypertable('eth.state_cids_i', 'block_number', chunk_time_interval => 32768);
SELECT create_distributed_hypertable('eth.storage_cids_i', 'block_number', chunk_time_interval => 32768);
SELECT create_distributed_hypertable('eth.state_accounts_i', 'block_number', chunk_time_interval => 32768);
SELECT create_distributed_hypertable('eth.access_list_elements_i', 'block_number', chunk_time_interval => 32768);
SELECT create_distributed_hypertable('eth.log_cids_i', 'block_number', chunk_time_interval => 32768);
-- migrate data
INSERT INTO eth.log_cids_i (SELECT * FROM eth.log_cids);
INSERT INTO eth.access_list_elements_i (SELECT * FROM eth.access_list_elements);
INSERT INTO eth.state_accounts_i (SELECT * FROM eth.state_accounts);
INSERT INTO eth.storage_cids_i (SELECT * FROM eth.storage_cids);
INSERT INTO eth.state_cids_i (SELECT * FROM eth.state_cids);
INSERT INTO eth.receipt_cids_i (SELECT * FROM eth.receipt_cids);
INSERT INTO eth.transaction_cids_i (SELECT * FROM eth.transaction_cids);
INSERT INTO eth.uncle_cids_i (SELECT * FROM eth.uncle_cids);
INSERT INTO eth.header_cids_i (SELECT * FROM eth.header_cids);
INSERT INTO public.blocks_i (SELECT * FROM public.blocks);
-- drop tables
DROP TABLE eth.log_cids;
DROP TABLE eth.access_list_elements;
DROP TABLE eth.state_accounts;
DROP TABLE eth.storage_cids;
DROP TABLE eth.state_cids;
DROP TABLE eth.receipt_cids;
DROP TABLE eth.transaction_cids;
DROP TABLE eth.uncle_cids;
DROP TABLE eth.header_cids;
DROP TABLE public.blocks;
-- rename distributed hypertables
ALTER TABLE eth.log_cids_i RENAME TO log_cids;
ALTER TABLE eth.access_list_elements_i RENAME TO access_list_elements;
ALTER TABLE eth.state_accounts_i RENAME TO state_accounts;
ALTER TABLE eth.storage_cids_i RENAME TO storage_cids;
ALTER TABLE eth.state_cids_i RENAME TO state_cids;
ALTER TABLE eth.receipt_cids_i RENAME TO receipt_cids;
ALTER TABLE eth.transaction_cids_i RENAME TO transaction_cids;
ALTER TABLE eth.uncle_cids_i RENAME TO uncle_cids;
ALTER TABLE eth.header_cids_i RENAME TO header_cids;
ALTER TABLE public.blocks_i RENAME TO blocks;
CALL distributed_exec('ALTER TABLE eth.log_cids_i RENAME TO log_cids');
CALL distributed_exec('ALTER TABLE eth.access_list_elements_i RENAME TO access_list_elements');
CALL distributed_exec('ALTER TABLE eth.state_accounts_i RENAME TO state_accounts');
CALL distributed_exec('ALTER TABLE eth.storage_cids_i RENAME TO storage_cids');
CALL distributed_exec('ALTER TABLE eth.state_cids_i RENAME TO state_cids');
CALL distributed_exec('ALTER TABLE eth.receipt_cids_i RENAME TO receipt_cids');
CALL distributed_exec('ALTER TABLE eth.transaction_cids_i RENAME TO transaction_cids');
CALL distributed_exec('ALTER TABLE eth.uncle_cids_i RENAME TO uncle_cids');
CALL distributed_exec('ALTER TABLE eth.header_cids_i RENAME TO header_cids');
CALL distributed_exec('ALTER TABLE public.blocks_i RENAME TO blocks');
-- update version
INSERT INTO public.db_version (singleton, version) VALUES (true, 'v4.0.00-dh')
ON CONFLICT (singleton) DO UPDATE SET (version, tstamp) = ('v4.0.0-dh', NOW());
-- +goose Down
INSERT INTO public.db_version (singleton, version) VALUES (true, 'v4.0.0')
ON CONFLICT (singleton) DO UPDATE SET (version, tstamp) = ('v4.0.0', NOW());
-- reversing conversion to hypertables requires migrating all data from every chunk back to a single table
-- create new regular tables
CREATE TABLE eth.log_cids_i (LIKE eth.log_cids INCLUDING ALL);
CREATE TABLE eth.access_list_elements_i (LIKE eth.access_list_elements INCLUDING ALL);
CREATE TABLE eth.state_accounts_i (LIKE eth.state_accounts INCLUDING ALL);
CREATE TABLE eth.storage_cids_i (LIKE eth.storage_cids INCLUDING ALL);
CREATE TABLE eth.state_cids_i (LIKE eth.state_cids INCLUDING ALL);
CREATE TABLE eth.receipt_cids_i (LIKE eth.receipt_cids INCLUDING ALL);
CREATE TABLE eth.transaction_cids_i (LIKE eth.transaction_cids INCLUDING ALL);
CREATE TABLE eth.uncle_cids_i (LIKE eth.uncle_cids INCLUDING ALL);
CREATE TABLE eth.header_cids_i (LIKE eth.header_cids INCLUDING ALL);
CREATE TABLE public.blocks_i (LIKE public.blocks INCLUDING ALL);
-- migrate data
INSERT INTO eth.log_cids_i (SELECT * FROM eth.log_cids);
INSERT INTO eth.access_list_elements_i (SELECT * FROM eth.access_list_elements);
INSERT INTO eth.state_accounts_i (SELECT * FROM eth.state_accounts);
INSERT INTO eth.storage_cids_i (SELECT * FROM eth.storage_cids);
INSERT INTO eth.state_cids_i (SELECT * FROM eth.state_cids);
INSERT INTO eth.receipt_cids_i (SELECT * FROM eth.receipt_cids);
INSERT INTO eth.transaction_cids_i (SELECT * FROM eth.transaction_cids);
INSERT INTO eth.uncle_cids_i (SELECT * FROM eth.uncle_cids);
INSERT INTO eth.header_cids_i (SELECT * FROM eth.header_cids);
INSERT INTO public.blocks_i (SELECT * FROM public.blocks);
-- drop distributed hypertables
DROP TABLE eth.log_cids;
DROP TABLE eth.access_list_elements;
DROP TABLE eth.state_accounts;
DROP TABLE eth.storage_cids;
DROP TABLE eth.state_cids;
DROP TABLE eth.receipt_cids;
DROP TABLE eth.transaction_cids;
DROP TABLE eth.uncle_cids;
DROP TABLE eth.header_cids;
DROP TABLE public.blocks;
-- rename hypertable tables
ALTER TABLE eth.log_cids_i RENAME TO log_cids;
ALTER TABLE eth.access_list_elements_i RENAME TO access_list_elements;
ALTER TABLE eth.state_accounts_i RENAME TO state_accounts;
ALTER TABLE eth.storage_cids_i RENAME TO storage_cids;
ALTER TABLE eth.state_cids_i RENAME TO state_cids;
ALTER TABLE eth.receipt_cids_i RENAME TO receipt_cids;
ALTER TABLE eth.transaction_cids_i RENAME TO transaction_cids;
ALTER TABLE eth.uncle_cids_i RENAME TO uncle_cids;
ALTER TABLE eth.header_cids_i RENAME TO header_cids;
ALTER TABLE public.blocks_i RENAME TO blocks;

View File

@ -1,5 +0,0 @@
-- +goose Up
-- to be generated by a script
-- +goose Down
-- to be generated by a script

View File

@ -1,27 +0,0 @@
-- +goose Up
-- to be generated by a script
SELECT set_replication_factor('public.blocks', 3);
SELECT set_replication_factor('eth.header_cids', 3);
SELECT set_replication_factor('eth.uncle_cids', 3);
SELECT set_replication_factor('eth.transaction_cids', 3);
SELECT set_replication_factor('eth.receipt_cids', 3);
SELECT set_replication_factor('eth.state_cids', 3);
SELECT set_replication_factor('eth.storage_cids', 3);
SELECT set_replication_factor('eth.state_accounts', 3);
SELECT set_replication_factor('eth.access_list_elements', 3);
SELECT set_replication_factor('eth.log_cids', 3);
-- +goose Down
-- to be generated by a script
SELECT set_replication_factor('public.blocks', 1);
SELECT set_replication_factor('eth.header_cids', 1);
SELECT set_replication_factor('eth.uncle_cids', 1);
SELECT set_replication_factor('eth.transaction_cids', 1);
SELECT set_replication_factor('eth.receipt_cids', 1);
SELECT set_replication_factor('eth.state_cids', 1);
SELECT set_replication_factor('eth.storage_cids', 1);
SELECT set_replication_factor('eth.state_accounts', 1);
SELECT set_replication_factor('eth.access_list_elements', 1);
SELECT set_replication_factor('eth.log_cids', 1);

View File

@ -5,58 +5,10 @@ services:
image: timescale/timescaledb:latest-pg14
restart: always
container_name: access-node
depends_on:
- data-node-1
- data-node-2
- data-node-3
command: ["postgres", "-c", "log_statement=all"]
environment:
POSTGRES_USER: "postgres"
POSTGRES_USER: "vdbm"
POSTGRES_DB: "vulcanize_testing_v4"
POSTGRES_PASSWORD: "password"
ports:
- "127.0.0.1:8066:5432"
volumes:
- ./scripts/init-access-node.sh:/docker-entrypoint-initdb.d/init-access-node.sh
data-node-1:
image: timescale/timescaledb:latest-pg14
restart: unless-stopped
container_name: data-node-1
command: ["postgres", "-c", "log_statement=all"]
environment:
POSTGRES_USER: "postgres"
POSTGRES_DB: "vulcanize_testing_v4"
POSTGRES_PASSWORD: "password"
ports:
- "127.0.0.1:8067:5432"
volumes:
- ./scripts/init-data-node.sh:/docker-entrypoint-initdb.d/init-data-node.sh
data-node-2:
image: timescale/timescaledb:latest-pg14
restart: unless-stopped
container_name: data-node-2
command: ["postgres", "-c", "log_statement=all"]
environment:
POSTGRES_USER: "postgres"
POSTGRES_DB: "vulcanize_testing_v4"
POSTGRES_PASSWORD: "password"
ports:
- "127.0.0.1:8068:5432"
volumes:
- ./scripts/init-data-node.sh:/docker-entrypoint-initdb.d/init-data-node.sh
data-node-3:
image: timescale/timescaledb:latest-pg14
restart: unless-stopped
container_name: data-node-3
command: ["postgres", "-c", "log_statement=all"]
environment:
POSTGRES_USER: "postgres"
POSTGRES_DB: "vulcanize_testing_v4"
POSTGRES_PASSWORD: "password"
ports:
- "127.0.0.1:8069:5432"
volumes:
- ./scripts/init-data-node.sh:/docker-entrypoint-initdb.d/init-data-node.sh

View File

@ -1,138 +0,0 @@
#!/bin/bash
# Guards
if [ -z "$1" ] || [ -z "$2" ] || [ -z "$3" ] || [ -z "$4" ] || [ -z "$5" ]
then
echo "Env variables not provided"
echo "Usage: ./add_nodes.sh <MIGRATION_FILE_PATH:string> <NODE_NAME:string> <NODE_HOST:string> <NODE_PORT:numeric/string> <NODE_DATABASE:string> <EXECUTE_SQL:bool>"
echo "Only <EXECUTE_SQL> is optional"
exit 1
fi
if [ -z "$6" ]
then
echo "EXECUTE_SQL not set, will not run statements against an access server"
else
echo "EXECUTE_SQL is set, will run stataments against an access server"
echo "Note: this mode is not recommended except in the case when the migration has already been applied with previous
nodes and we need/want to add more while still recording them in the existing migration (and adding their Down statements to said migration)"
echo "Expected environment variables:"
echo "DATABASE_HOSTNAME, DATABASE_NAME, DATABASE_PORT, DATABASE_USER"
echo "For now, DATABASE_PASSWORD will be prompted for on each psql call"
fi
# Remote DB node info
export MIGRATION_FILE_PATH=$1
export NODE_NAME=$2
export NODE_HOST=$3
export NODE_PORT=$4
export NODE_DATABASE=$5
printf "Enter the ${NODE_HOST} database password:\n"
stty -echo
read NODE_PASSWORD
stty echo
export NODE_PASSWORD
if ! [ -z "$6" ]
then
# Access DB info
echo "heeeeey"
export DATABASE_HOSTNAME=localhost
export DATABASE_PORT=5432
export DATABASE_USER=vdbm
export DATABASE_NAME=vulcanize_db
fi
# Array of distributed hypertable names
declare -a tables_names=("public.blocks" "eth.header_cid" "eth.uncle_cids" "eth.transaction_cids"
"eth.receipt_cids" "eth.state_cid" "eth.storage_cids" "eth.state_accounts"
"eth.access_list_elements" "eth.log_cids"
)
# Array to append Up statements to for later (optional) execution
declare -a up_stmts=()
echo "Writing Up and Down statements to provided migration file at ${migration_file_path}"
# Create add node statement
up_add_pg_str="SELECT add_data_node('${NODE_NAME}', host => '${NODE_HOST}', port => ${NODE_PORT}, database => '${NODE_DATABASE}', password => '${NODE_PASSWORD}');"
up_stmts+=(${up_add_pg_str})
# Insert at the 3rd line of the file
sed -i.bak '3 i\
'"${up_add_pg_str}"'
' "${MIGRATION_FILE_PATH}"
# Check for error
if [[ $? -eq 0 ]]; then
echo "Wrote Up add node statement ${up_add_pg_str}"
else
echo "Could not write Up add node statement ${up_add_pg_str}. Is the migration file path correct?"
exit 1
fi
# Create attach node statements
for table_name in "${tables_names[@]}"
do
up_attach_pg_str="SELECT attach_data_node('${NODE_NAME}', '${table_name}', if_not_attached => true);"
up_stmts+=(${up_attach_pg_str})
# Insert at the 4th line of the file
sed -i.bak '4 i\
'"${up_attach_pg_str}"'
' "${MIGRATION_FILE_PATH}"
# Check for error
if [[ $? -eq 0 ]]; then
echo "Wrote Up attach node statement ${up_attach_pg_str}"
else
echo "Could not write Up attach node statement ${up_attach_pg_str}. Is the migration file path correct?"
exit 1
fi
done
## Create detach and remove node statement
down_attach_pg_str="SELECT detach_data_node('${NODE_NAME}', force => true, if_attached = true);"
down_add_pg_str="SELECT delete_data_node('${NODE_NAME}', force => true, if_attached => true);"
# Append them at the last line in the file
sed -i.bak '$ a\
'"${down_attach_pg_str}"'
' "${MIGRATION_FILE_PATH}"
# Check for error
if [[ $? -eq 0 ]]; then
echo "Wrote Down attach node statement ${down_attach_pg_str}"
else
echo "Could not write Down attach node statement ${down_attach_pg_str}. Is the migration file path correct?"
exit 1
fi
# Append them at the last line in the file
sed -i.bak '$ a\
'"${down_add_pg_str}"'
' "${MIGRATION_FILE_PATH}"
# Check for error
if [[ $? -eq 0 ]]; then
echo "Wrote Down add node statement ${down_add_pg_str}"
else
echo "Could not write Down add node statement ${down_add_pg_str}. Is the migration file path correct?"
exit 1
fi
# Execute Up statements on the server if we are in that mode
if [ -z "$6" ]
then
echo "Done!"
exit 0
fi
echo "Executing Up statements against provided server"
for up_stmt in "${up_stmts[@]}"
do
psql -c '\x' -c "${up_stmt}" -h $DATABASE_HOSTNAME -p $DATABASE_PORT -U $DATABASE_USER -d $DATABASE_NAME -W
# Check for error
if [[ $? -eq 0 ]]; then
echo "Executed Up statement ${up_stmt}}"
else
echo "Could not execute Up statement ${up_stmt}. Is the migration file path correct?"
exit 1
fi
done
echo "Done!"
exit 0

View File

@ -1,29 +0,0 @@
#!/bin/sh
set -e
# https://docs.timescale.com/timescaledb/latest/how-to-guides/multinode-timescaledb/multinode-config/
# To achieve good query performance you need to enable partition-wise aggregation on the access node. This pushes down aggregation queries to the data nodes.
# https://www.postgresql.org/docs/12/runtime-config-query.html#enable_partitionwise_aggregate
sed -ri "s!^#?(enable_partitionwise_aggregate)\s*=.*!\1 = on!" /var/lib/postgresql/data/postgresql.conf
grep "enable_partitionwise_aggregate" /var/lib/postgresql/data/postgresql.conf
# JIT should be set to off on the access node as JIT currently doesn't work well with distributed queries.
# https://www.postgresql.org/docs/12/runtime-config-query.html#jit
sed -ri "s!^#?(jit)\s*=.*!\1 = off!" /var/lib/postgresql/data/postgresql.conf
grep "jit" /var/lib/postgresql/data/postgresql.conf
# https://docs.timescale.com/timescaledb/latest/how-to-guides/multinode-timescaledb/multinode-auth/
# https://docs.timescale.com/timescaledb/latest/how-to-guides/multinode-timescaledb/multinode-auth/#password-authentication
# Set password_encryption = 'scram-sha-256' in postgresql.conf on the access node.
sed -ri "s!^#?(password_encryption)\s*=.*!\1 = 'scram-sha-256'!" /var/lib/postgresql/data/postgresql.conf
grep "password_encryption" /var/lib/postgresql/data/postgresql.conf
# Append to data/passfile *:*:*:ROLE:ROLE_PASSWORD
# This file stores the passwords for each role that the access node connects to on the data nodes.
echo "*:*:*:postgres:password">>/var/lib/postgresql/data/passfile
chmod 0600 /var/lib/postgresql/data/passfile
# Add "host all all ACCESS_NODE_IP scram-sha-256" pg_hba.conf on the data nodes.
# Skipped. Using default "host all all all scram-sha-256" for now.

View File

@ -1,27 +0,0 @@
#!/bin/sh
set -e
# https://docs.timescale.com/timescaledb/latest/how-to-guides/configuration/timescaledb-config/#timescaledb-last-tuned-string
# https://docs.timescale.com/timescaledb/latest/how-to-guides/multi-node-setup/required-configuration/
# It is necessary to change the parameter max_prepared_transactions to a non-zero value ('150' is recommended).
# https://www.postgresql.org/docs/12/runtime-config-resource.html#max_prepared_transactions
sed -ri "s!^#?(max_prepared_transactions)\s*=.*!\1 = 150!" /var/lib/postgresql/data/postgresql.conf
grep "max_prepared_transactions" /var/lib/postgresql/data/postgresql.conf
# Statement timeout should be disabled on the data nodes and managed through the access node configuration if desired.
# https://www.postgresql.org/docs/12/runtime-config-client.html#statement_timeout
sed -ri "s!^#?(statement_timeout)\s*=.*!\1 = 0!" /var/lib/postgresql/data/postgresql.conf
grep "statement_timeout" /var/lib/postgresql/data/postgresql.conf
# On the data nodes, set the wal_level to logical or higher to move or copy chunks between data nodes.
# https://www.postgresql.org/docs/14/runtime-config-wal.html
sed -ri "s!^#?(wal_level)\s*=.*!\1 = 'logical'!" /var/lib/postgresql/data/postgresql.conf
grep "wal_level" /var/lib/postgresql/data/postgresql.conf
# https://docs.timescale.com/timescaledb/latest/how-to-guides/multinode-timescaledb/multinode-auth/
# https://docs.timescale.com/timescaledb/latest/how-to-guides/multinode-timescaledb/multinode-auth/#password-authentication
# Set password_encryption = 'scram-sha-256' in postgresql.conf on the data node.
sed -ri "s!^#?(password_encryption)\s*=.*!\1 = 'scram-sha-256'!" /var/lib/postgresql/data/postgresql.conf
grep "password_encryption" /var/lib/postgresql/data/postgresql.conf

View File

@ -7,9 +7,8 @@ VDB_PG_CONNECT=postgresql://$DATABASE_USER:$DATABASE_PASSWORD@$DATABASE_HOSTNAME
# Run the DB migrations
echo "Connecting with: $VDB_PG_CONNECT"
sleep 15
echo "Running database migrations"
./goose -dir migrations/vulcanizedb postgres "$VDB_PG_CONNECT" up-to 24
./goose -dir migrations/vulcanizedb postgres "$VDB_PG_CONNECT" up
# If the db migrations ran without err
if [[ $? -eq 0 ]]; then