Postgres Pro
Downloads
Home

Postgres Pro Enterprise in Yandex.Cloud

Postgres Pro Enterprise in Yandex.Cloud Quick Start Guide

Internet access and valid Yandex.Cloud account are required to use Postgres Pro Enterprise 10/11/12/13/14 database in Yandex.Cloud.

Postgres Pro Enterprise 10/11/12/13/14 virtual machine (VM) image is available in Yandex.Cloud Marketplace.

Software required for installation:

  • 'yc' as part of Yandex.Cloud CLI for cloud management
  • 'psql' or 'Pgadmin' for database connection

Yandex.Cloud Console https://console.cloud.yandex.ru can be used as well.

Yandex.Cloud CLI installation guide: https://cloud.yandex.ru/docs/cli/quickstart

Yandex.Cloud CLI reference guide: https://cloud.yandex.ru/docs/cli/

 

Connection to Yandex.Cloud and environment check

  • Connect to Yandex.Cloud and configure environment with:
yc init
  • Verify 'yc' version (should be the latest available):
yc version
  • A list of regions and zones for VM:
yc compute zone list
yc config set compute-default-zone ru-central1-a
yc config get compute-default-zone

‘ru-central1’ region and 'ru-central1-a' zone will be used further.

  • A list of available disk types:
yc compute disk-type list

‘network-hdd’ disk type will be used further.

 

VM creation

  • Create network infrastructure for VM:
NET=my-vpc

yc vpc network create \
--name $NET \
--description "$NET" \
--async

yc vpc network list

yc vpc subnet create \
--name $NET-subnet \
--range 192.168.0.0/24 \
--network-name $NET \
--description "$NET-subnet" \
--async

yc vpc subnet list
  • Create a pair of private/public ssh-keys in ~/.ssh directory for VM connection:
ssh-keygen -t rsa -b 2048
  • Use the following commands to create VM from the latest Postgres Pro Enterprise 14 image version available in Yandex.Cloud Marketplace:
IMAGE_FAMILY=postgreprof-postgres-pro-enterprise-database-14

VM_NAME=myvm-ent-xx

yc compute instance create \
--name $VM_NAME \
--hostname $VM_NAME \
--cores 2 \
--memory 2 \
--create-boot-disk name=$VM_NAME-osdisk,image-folder-id=standard-images,image-family=$IMAGE_FAMILY \
--network-interface subnet-name=$NET-subnet,nat-ip-version=ipv4 \
--ssh-key ~/.ssh/id_rsa.pub \
--async

yc compute instances show $VM_NAME

Replace ‘xx’ by '01', '02', '03’ and so on.

 

Connection to VM

  • Connect to VM with private ssh-key using VM's public IP-address ($VM_IP_ADDRESS):
yc compute instances list

VM_IP_ADDRESS=$(yc compute instance show --name $VM_NAME | grep -E ' +address' | tail -n 1 | awk '{print $2}')
echo $VM_IP_ADDRESS

ssh yc-user@$VM_IP_ADDRESS

 

Postgres Pro database service status

  • Verify Postgres Pro database service status: 
sudo systemctl -l status postgrespro-ent-14.service
  • To stop/start Postgres Pro database service use the following commands: 
sudo systemctl stop postgrespro-ent-14.service
sudo systemctl start postgrespro-ent-14.service

 

Connection to Postgres Pro database

  • Switch to ‘postgres’ account:
sudo su - postgres
  • To connect to Postgres Pro database use the following command:
psql
  • To exit from ‘psql’ use the following command:
\q
  • To return to 'yc' interface run 'exit' command twice

 

External connection to VM

  • TCP-port 5433 has to be opened for external connection to Postgres Pro database: 
ssh yc-user@$VM_IP_ADDRESS

sudo firewall-cmd --zone=public --add-port=5433/tcp
sudo firewall-cmd --zone=public --permanent --add-port=5433/tcp
sudo firewall-cmd --zone=public --list-ports

sudo systemctl restart firewalld.service
sudo systemctl -l status firewalld.service

exit
  • TCP-ports 80 and 443 have to be opened for external connection to database monitoring server: 
ssh yc-user@$VM_IP_ADDRESS

sudo firewall-cmd --zone=public --add-port=80/tcp
sudo firewall-cmd --zone=public --permanent --add-port=80/tcp
sudo firewall-cmd --zone=public --add-port=443/tcp
sudo firewall-cmd --zone=public --permanent --add-port=443/tcp
sudo firewall-cmd --zone=public --list-ports

sudo systemctl restart firewalld.service
sudo systemctl -l status firewalld.service

exit

 

External connection to Postgres Pro database

  • For external connection to Postgres Pro database set up 'postgres' user password:
ssh yc-user@$VM_IP_ADDRESS

sudo su - postgres
psql -c "alter user postgres with encrypted password 'YOUR_POSTGRES_USER_PASSWORD'"
exit

exit
  • For external connection to Postgres Pro database using ‘psql’ utility use the following command:
psql --host=$VM_IP_ADDRESS --port=5433 --username=postgres --dbname=postgres
  • For external connection to Postgres Pro database using ‘Pgadmin’ utility configure the following server settings in ‘Pgadmin’ menu:
    • ‘mydb-xx’ for ‘Name’
    • ‘$VM_IP_ADDRESS’ for ‘Host’
    • ‘5433’ for ‘Port’
    • ‘postgres’ for ‘Maintenance DB’
    • ‘postgres’ for ‘Username’

Replace ‘xx’ by '01', '02', '03’ and so on.

Replace '$VM_IP_ADDRESS' by VM's external IP-address.

 

External connection to database monitoring server

  • For connection to database monitoring server set up 'Admin' user password:
ssh yc-user@$VM_IP_ADDRESS

sudo su - postgres
source .pgsql_profile
psql --dbname=zabbix --username=zabbix -c "update users set passwd=md5('YOUR_ZABBIX_ADMIN_PASSWORD') where alias='Admin'"
exit

exit
  • External connection to database monitoring server is via the following link:

https://$VM_IP_ADDRESS/zabbix

Replace ‘$VM_IP_ADDRESS’ by VM's external IP-address.

 

VM configuration change

Let's have some examples of VM configuration change:

1) Change VM type to increase VM computing power

  • $PGDATA/postgresql.tune.lock file can be deleted in order to automatically adjust Postgres Pro database parameters values before increasing VM size:
ssh yc-user@$VM_IP_ADDRESS

sudo su - postgres
cp $PGDATA/postgresql.auto.conf $PGDATA/postgresql.auto.conf.ORIG
rm $PGDATA/postgresql.tune.lock
exit

exit
  • To change VM type use the following commands:
yc compute instances show $VM_NAME

yc compute instances stop $VM_NAME

yc compute instances update $VM_NAME \
--cores 4 \
--memory 4

yc compute instances start $VM_NAME

yc compute instances list

VM_IP_ADDRESS=$(yc compute instance show --name $VM_NAME | grep -E ' +address' | tail -n 1 | awk '{print $2}')
echo $VM_IP_ADDRESS

ssh yc-user@$VM_IP_ADDRESS

cat /proc/cpuinfo
cat /proc/meminfo

exit

2) Increase OS-disk size up to 80 GB

  • Obtain OS-disk size details: 
yc compute disks show $VM_NAME-osdisk
  • Stop VM temporarily: 
yc compute instances stop $VM_NAME
  • Increase OS-disk size: 
yc compute disks update $VM_NAME-osdisk \
--size 80
  • Verify new OS-disk size: 
yc compute disks show $VM_NAME-osdisk
  • Start VM: 
yc compute instances start $VM_NAME
  • Connect to VM: 
yc compute instances list

VM_IP_ADDRESS=$(yc compute instance show --name $VM_NAME | grep -E ' +address' | tail -n 1 | awk '{print $2}')
echo $VM_IP_ADDRESS

ssh yc-user@$VM_IP_ADDRESS

df -h /

exit

3) Use dedicated datadisk for Postgres Pro database files to improve database performance

  • Let's create new 200 GB datadisk and attach it to VM:
yc compute disks create \
--name $VM_NAME-datadisk \
--size 200 \
--type network-hdd \
--async

yc compute disks list

yc compute instances attach-disk \
--name $VM_NAME \
--disk-name $VM_NAME-datadisk \
--async

yc compute disks list
  • Connect to VM:
ssh yc-user@$VM_IP_ADDRESS
  • Stop Postgres Pro database service and verify its status:
sudo systemctl stop postgrespro-ent-14.service
sudo systemctl -l status postgrespro-ent-14.service
  • Create new filesystem mountpoint:
sudo mkdir /PGDATA
  • Use 'fdisk' utility to find out datadisk device name (in this case it is '/dev/vdb'):
sudo fdisk -l
  • Create datadisk single partition of maximum size, create new filesystem on top of it:
(echo n; echo p; echo 1; echo ; echo ; echo w) | sudo fdisk /dev/vdb
sudo mkfs -t ext4 /dev/vdb1
  • Amend /etc/fstab file for new filesystem automount and mount it:
sudo sh -c "echo '`sudo blkid -o export /dev/vdb1 | grep UUID` /PGDATA ext4 defaults,nofail,barrier=0 1 2' >> /etc/fstab"
sudo mount /PGDATA
  • Create 'data' directory on the new filesystem and set its permissions:
sudo mkdir /PGDATA/data
sudo chown postgres:postgres /PGDATA/data
sudo chmod 0700 /PGDATA/data
  • Switch to ‘postgres’ account and move Postgres Pro database files to the new filesystem:
sudo su - postgres
mv /var/lib/pgpro/ent-14/data/* /PGDATA/data; rmdir /var/lib/pgpro/ent-14/data; ln -s /PGDATA/data /var/lib/pgpro/ent-14/data
exit
  • Start Postgres Pro database service and verify its status:
sudo systemctl start postgrespro-ent-14.service
sudo systemctl -l status postgrespro-ent-14.service
  • Restart VM, check filesystem automount and verify Postgres Pro database service status:
sudo reboot

ssh yc-user@$VM_IP_ADDRESS

sudo mount | grep /PGDATA
sudo df -h /PGDATA

sudo systemctl -l status postgrespro-ent-14.service

Replace ‘xx’ by '01', '02', '03’ and so on.

4) Database service auto restart in case of database failure

  • Edit database service systemd file and restart database service:
sudo sed -i '/KillSignal=/a Restart=on-failure' /usr/lib/systemd/system/postgrespro-ent-14.service
sudo systemctl daemon-reload
sudo systemctl restart postgrespro-ent-14.service
sudo systemctl -l status postgrespro-ent-14.service

 

VM backup/restore

 

Postgres Pro Enterprise Multimaster

Multimaster is a Postgres Pro Enterprise extension with a set of patches that turns Postgres Pro Enterprise into a synchronous shared-nothing cluster to provide Online Transaction Processing (OLTP) scalability for read transactions and high availability with automatic disaster recovery.

You can find more information about Postgres Pro Enterprise Multimaster at:

https://postgrespro.com/docs/enterprise/14/multimaster

Let's proceed with 3-node Multimaster installation and configuration. Assume Multimaster node names are 'myvm-ent-01', 'myvm-ent-02' and 'myvm-ent-03'.

  • Create three VMs:
for i in `seq 1 3`; do
yc compute instance create \
--name myvm-ent-0$i \
--hostname myvm-ent-0$i \
--cores 2 \
--memory 4 \
--create-boot-disk name=myvm-ent-0$i-osdisk,image-folder-id=standard-images,image-family=$IMAGE_FAMILY \
--network-interface subnet-name=$NET-subnet,nat-ip-version=ipv4,ipv4-address=192.168.0.10$i \
--ssh-key ~/.ssh/id_rsa.pub \
--async
done

Private IP-addresses assignment: 192.168.0.101 (myvm-ent-01), 192.168.0.102 (myvm-ent-02), 192.168.0.103 (myvm-ent-03).

  • Now connect to each of three VMs:
for i in `seq 1 3`; do
export VM_IP_ADDRESS_0$i=$(yc compute instance show --name myvm-ent-0$i | grep -E ' +address' | tail -n 1 | awk '{print $2}')
done

ssh yc-user@$VM_IP_ADDRESS_01
ssh yc-user@$VM_IP_ADDRESS_02
ssh yc-user@$VM_IP_ADDRESS_03
  • and run the same set of commands on all of them:
for i in `seq 1 3`; do
sudo sh -c "echo '192.168.0.10$i myvm-ent-0$i' >> /etc/hosts"
done
  • Configure replicated 'mydb' database:
sudo su - postgres
psql -c "create user myuser with superuser encrypted password 'myuserpassword'"
psql --username=myuser -c "create database mydb"
sed -i 's/PGDATABASE=postgres/PGDATABASE=mydb/' .pgpro_profile
sed -i 's/PGUSER=postgres/PGUSER=myuser/' .pgpro_profile
source .pgpro_profile

for i in `seq 1 3`; do
echo "hostssl replication myuser myvm-ent-0$i md5" >> $PGDATA/pg_hba.conf
echo "myvm-ent-0$i:5433:mydb:myuser:myuserpassword" >> ~/.pgpass
done
chmod 0600 ~/.pgpass
pg_ctl reload

echo "" >> $PGDATA/postgresql.conf
echo "" >> $PGDATA/postgresql.conf
echo "#------------------------------------------------------------------------------" >> $PGDATA/postgresql.conf
echo "# MULTIMASTER SETTINGS" >> $PGDATA/postgresql.conf
echo "#------------------------------------------------------------------------------" >> $PGDATA/postgresql.conf
echo "" >> $PGDATA/postgresql.conf
echo "multimaster.max_nodes = 3" >> $PGDATA/postgresql.conf
echo "" >> $PGDATA/postgresql.conf
echo "" >> $PGDATA/postgresql.conf

psql -c "alter system set default_transaction_isolation to 'read committed'"
psql -c "alter system set wal_level to logical"
psql -c "alter system set max_connections to 100"
psql -c "alter system set max_prepared_transactions to 300"
psql -c "alter system set max_wal_senders to 10"
psql -c "alter system set max_replication_slots to 10"
psql -c "alter system set max_worker_processes to 250"
psql -c "alter system set shared_preload_libraries to multimaster,pg_stat_statements,pg_buffercache,pg_wait_sampling"
psql -c "alter system set wal_sender_timeout to 0"
exit
  • Configure mamonsu agent for 'mydb' database and restart mamonsu service:
sudo sed -i 's|user = mamonsu|user = myuser|' /etc/mamonsu/agent.conf
sudo sed -i 's|database = mamonsu|database = mydb|' /etc/mamonsu/agent.conf
sudo systemctl restart mamonsu.service
  • Restart Postgres Pro database service and verify its status:
sudo systemctl restart postgrespro-ent-14.service
sudo systemctl -l status postgrespro-ent-14.service

exit
  • Now connect to the first VM:
ssh yc-user@$VM_IP_ADDRESS_01
  • and create Multimaster extension:
sudo su - postgres
psql
create extension if not exists multimaster;
select mtm.init_cluster('dbname=mydb user=myuser host=myvm-ent-01 port=5433 sslmode=require','{"dbname=mydb user=myuser host=myvm-ent-02 port=5433 sslmode=require", "dbname=mydb user=myuser host=myvm-ent-03 port=5433 sslmode=require"}');
\q
  • Create other extensions required for mamonsu service:
psql -c "create extension if not exists pg_buffercache"
psql -c "create extension if not exists pg_stat_statements"
psql -c "create extension if not exists pg_wait_sampling"
  • Verify the extensions have been successfully created:
psql --host=myvm-ent-01 -c "select * from pg_extension"
psql --host=myvm-ent-02 -c "select * from pg_extension"
psql --host=myvm-ent-03 -c "select * from pg_extension"
  • Configure mamonsu for Multimaster:
mamonsu bootstrap --dbname mydb --username postgres --host 127.0.0.1 --port 5433 --mamonsu-username=myuser
psql --host=myvm-ent-01 -c "select mtm.make_table_local('mamonsu_config')"
psql --host=myvm-ent-01 -c "select mtm.make_table_local('mamonsu_timestamp_master_2_7_1')"
  • Use the following commands to monitor Multimaster status:
psql --host=myvm-ent-01 -x -c "select mtm.status()"
psql --host=myvm-ent-02 -x -c "select mtm.status()"
psql --host=myvm-ent-03 -x -c "select mtm.status()"

psql --host=myvm-ent-01 -x -c "select mtm.nodes()"
psql --host=myvm-ent-02 -x -c "select mtm.nodes()"
psql --host=myvm-ent-03 -x -c "select mtm.nodes()"
  • To return to 'yc' interface run 'exit' command twice

 

External connection to Postgres Pro Enterprise Multimaster database

 

 

Postgres Pro Enterprise Multimaster (2-node + referee node configuration)

If 3-node Multimaster is an overkill configuration (triple storage size for database), it's still possible to have 2-node Multimaster configuration either by nominating one of two nodes a 'major' node (multimaster.major_node=on) or by using a light-weight node as a referee instead of the third node.

The way to migrate from 3-node Multimaster to 2-node + referee node Multimaster is the following.

  • Exclude 'myvm-ent-03' node from Multimaster configuration, remove Multimster settings from 'myvm-ent-03' referee node and change Multimaster settings on 'myvm-ent-01' and 'myvm-ent-02' nodes:
sudo su - postgres
sed -i '/multimaster/d' $PGDATA/postgresql.conf
psql --host=myvm-ent-03 -c "alter system set shared_preload_libraries to pg_stat_statements,pg_buffercache,pg_wait_sampling"
psql --host=myvm-ent-01 -c "select mtm.drop_node(3)"
psql --host=myvm-ent-01 -x -c "select mtm.nodes()"
psql --host=myvm-ent-01 -c "alter system set multimaster.referee_connstring = 'dbname=mydb user=myuser host=myvm-ent-03 port=5433 sslmode=require'"
psql --host=myvm-ent-02 -c "alter system set multimaster.referee_connstring = 'dbname=mydb user=myuser host=myvm-ent-03 port=5433 sslmode=require'"
exit

sudo systemctl restart postgrespro-ent-14.service

sudo su - postgres
psql --host=myvm-ent-03 -c "drop extension multimaster"
psql --host=myvm-ent-03 -c "drop publication if exists multimaster"
psql --host=myvm-ent-03 -c "create extension referee"
exit

sudo systemctl restart postgrespro-ent-14.service
  • Apply new Multimster settings on 'myvm-ent-01' and 'myvm-ent-02' nodes:
sudo systemctl restart postgrespro-ent-14.service
  • Finally, drop replication slots and finalize settings on 'myvm-ent-03' referee node:
sudo su - postgres
sed -i '/^#/!d' $PGDATA/postgresql.auto.conf
echo "shared_preload_libraries = 'pg_stat_statements, pg_buffercache, pg_wait_sampling'" >> $PGDATA/postgresql.auto.conf
psql --host=myvm-ent-03 -c "select pg_drop_replication_slot('mtm_slot_1')"
psql --host=myvm-ent-03 -c "select pg_drop_replication_slot('mtm_filter_slot_1')"
psql --host=myvm-ent-03 -c "select pg_drop_replication_slot('mtm_slot_2')"
psql --host=myvm-ent-03 -c "select pg_drop_replication_slot('mtm_filter_slot_2')"
psql --host=myvm-ent-03 -c "select pg_replication_origin_drop('mtm_slot_1')"
psql --host=myvm-ent-03 -c "select pg_replication_origin_drop('mtm_slot_2')"
exit

sudo systemctl restart postgrespro-ent-14.service
  • Check the status of Multimaster on each node:
sudo su - postgres
psql --dbname=mydb --username=myuser --host=myvm-ent-01 --port=5433 -x -c "select mtm.nodes()"
psql --dbname=mydb --username=myuser --host=myvm-ent-02 --port=5433 -x -c "select mtm.nodes()"
psql --dbname=mydb --username=myuser --host=myvm-ent-03 --port=5433 -c "select * from referee.decision"
exit

 

Postgres Pro Enterprise CFS (compressed file system):

In order to use functionality of Postgres Pro Enterprise CFS (compressed file system) extension, proceed with the following commands.

  • Create OS filesystem directory for 'cfs_ts' tablespace and 'cfs_ts' tablespace in the database: 
sudo su - postgres
mkdir $PGDATA/../cfs_ts
chmod 0700 $PGDATA/../cfs_ts

psql -c "create tablespace cfs_ts location '/var/lib/pgpro/ent-14/cfs_ts' with (compression=true)"

exit
  • Make sure the 'cfs_ts' tablespace has been created with 'compression=true' option: 
sudo su - postgres
psql -c "select * from pg_tablespace"
exit

Use one of the following ways to utilise 'cfs_ts' tablespace for new database objects:

  • Upon database object creation: 
sudo su - postgres
psql -c "create table t1 (t int) tablespace cfs_ts"
psql -c "select tablename, tablespace from pg_tables where schemaname = 'public'"
exit
  • Setting default tablespace for current database connection: 
sudo su - postgres
psql
set default_tablespace=cfs_ts;
show default_tablespace;
create table t2 (t int);
select tablename, tablespace from pg_tables where schemaname = 'public';
\q
exit
  • Setting default tablespace for particular database: 
sudo su - postgres
psql --dbname=postgres -c "alter database mydb set tablespace cfs_ts"
psql -c "select datname, dattablespace from pg_database"
exit
  • Setting default tablespace for particular user/role: 
sudo su - postgres
psql --username=postgres -c "alter user myuser set default_tablespace to 'cfs_ts'"
psql -c "select usename, useconfig from pg_user"
psql -c "select rolname, rolconfig from pg_roles"
exit

Use one of the following ways to move existing database objects from one tablespace ('pg_default') to another ('cfs_ts'):

  • One by one:
sudo su - postgres
psql -c "create table t3 (t int)"
psql -c "alter table t3 set tablespace cfs_ts"
psql -c "select tablename, tablespace from pg_tables where schemaname = 'public'"
exit
  • All together:
sudo su - postgres
psql -c "alter table all in tablespace pg_default set tablespace cfs_ts"
psql -c "select tablename, tablespace from pg_tables where schemaname = 'public'"
exit

Depending on the data stored in tablespace 'cfs_ts', commpression ratio may vary.

 

Postgres Pro Enterprise 10/11/12/13/14 VM content:

Linux Debian 10 (64-bit)

  • OS-account - ‘postgres’
  • OS-account - ‘zabbix’
  • OS-account - ‘mamonsu’

OS-disk size - 10 GB

  • xfs filesystem ‘/’ (/dev/vda1) - 10 GB

Main database - Postgres Pro Enterprise

  • DB version: 10/11/12/13/14
  • TCP-port: 5433 (opened in OS-firewall settings)
  • configuration file: /var/lib/pgsql/.pgpro_profile
  • database account: ’postgres’

Database monitoring (server)

  • zabbix-server version: 4.x
  • TCP-ports: 80/443 (opened in OS-firewall settings)
  • account: ‘Admin’

Database monitoring (agent)

  • zabbix-agent version: 4.x
  • mamonsu-agent version: 2.x
  • configuration file: /etc/mamonsu/agent.conf

Auxiliary database PostgreSQL (as a zabbix-server database)

  • DB version: 10/11/12/13/14
  • TCP-port: 5432
  • configuration file: /var/lib/pgsql/.pgsql_profile
  • database account: 'postgres'

 

Documentation links