TiKV集群配置记录

环境：两台ubuntu 18.04 pc

　　tidb / pd: 192.168.1.150

　　tikv: 192.1681.1.151

主要参考https://pingcap.com/docs-cn/dev/how-to/deploy/orchestrated/ansible/

git clone https://github.com/pingcap/go-ycsb.git
git clone https://github.com/pingcap/tidb-ansible.git
vi hosts.ini
vi inventory.ini

ansible-playbook -i hosts.ini deploy_ntp.yml -u pingcap -b

ansible-playbook local_prepare.yml
ansible-playbook bootstrap.yml --extra-vars "dev_mode=True"
ansible-playbook deploy.yml
ansible-playbook start.yml

有几点补充：

1. 配置前准备：

　　0. 安装时就将用户名设为tidb，且计算机名要互不相同

　　1. 开启ssh，设置允许root ssh

　　2. 设置ext4分区格式（参考上述链接）

　　3. 安装1.12.2版本的go

　　4. 使用master版本的tidb-ansible。在inventory.ini中tidb_version使用默认值latest

2. 在ansible-playbook bootstrap.yml这里，加上如下参数来跳过配置自检：

　　ansible-playbook bootstrap.yml --extra-vars "dev_mode=True"

3. 单机部署时，如果ansible start阶段遇到以下错误：

TASK [wait for region replication complete] ****************************************************************************
FAILED - RETRYING: wait for region replication complete ( retries left).
FAILED - RETRYING: wait for region replication complete ( retries left).
FAILED - RETRYING: wait for region replication complete ( retries left).
FAILED - RETRYING: wait for region replication complete ( retries left).
FAILED - RETRYING: wait for region replication complete ( retries left).
FAILED - RETRYING: wait for region replication complete ( retries left).
FAILED - RETRYING: wait for region replication complete ( retries left).
FAILED - RETRYING: wait for region replication complete ( retries left).
FAILED - RETRYING: wait for region replication complete ( retries left).

tidb-ansible/conf/pd.yml里面 max-replicas 改成1，然后ansible unsafe_cleanup，重新deploy

Ref:

https://www.cnblogs.com/rongfengliang/p/8006165.html

https://blog.csdn.net/lockey23/article/details/78256143

-------------------------------------------------配置ycsb-------------------------------------------------

go-ycsb是一个在tikv上运行benchmark的工具

首先把https://github.com/pingcap/go-ycsb给git clone到本地，然后make

tidb@pd1:~/Desktop/go-ycsb$ ./bin/go-ycsb load tikv -P workloads/workloada -p tikv.pd=192.168.1.150:2379

tidb@pd1:~/Desktop/go-ycsb$ ./bin/go-ycsb run tikv -P workloads/workloada -p tikv.pd=192.168.1.150:2379

注：如果报 ERRO[0000] batch commands send error: EOF 错误可以忽略。[ https://github.com/tikv/client-go/issues/19 ]

Ref:

https://blog.csdn.net/qq_32523587/article/details/82146789

https://www.jianshu.com/p/73d22befe47d