Clickhouse For CentOs
JDK
Using JDK 11 as well as enabling G1 garbage collector is recommended. Comparing to JDK 8, the performance of handling jobs under multi-threads will be improved more than 30%.
Clickhouse
1. Download
2. Install
3. Memory Settings
We will need to limit the clickhouse service If the client server has a memory less than 32GB.
clickhouse memory settings
In /etc/clickhouse-server/config.xml
set max_server_memory_usage_to_ram_ratio
to 0.125 so that 32G*0.125 will be 4G
Update /etc/clickhouse-server/users.xml
and set max_memory_usage
to 4000000000, max_bytes_before_external_group_by
to 2000000000. You may need to create max_bytes_before_external_group_by
if it isn't present.
Restart the Clickhouse service: sudo /etc/init.d/clickhouse-server restart
Sqlservice memory settings
With this config, the Simple mode will use 22G memory and the Rugular mode will be in 18G memory.
4. Set Clickhouse default password
Set Clickhouse password to sqlflow@gudu
If mistakenly set to another password, you will need:
delete
/etc/clickhouse-server/users.d/default-password.xml
update
/etc/clickhouse-server/users.xml
, set password to sqlflow@gudurestart Clickhouse,
sudo /etc/init.d/clickhouse-server restart
5. Init Clickhouse
If you got
Then you might need to check for your OS version and the Clickhouse is not properly installed. Check Ubuntu/Debian/RHEL Installation.
If your clickhouse is properly installed but you still got errors, please check /var/log/clickhouse-sever
for error logs.
6. Start SQLFlow
Last updated