Monitoring and Management of Percona XtraDB

From ppwiki
Jump to: navigation, search

For this tutorial, we will install Percona Monitoring and Management server (PMM server) on a separate node and pmm-client on each database node

Prerequisites

We have three (3) different ways of running PMM server 1- Running PMM server via Docker 2- Running PMM server as a Virtual Appliance 3- Running PMM server Using AWS Marketplace

For the purpose of this tutorial, we we be running PMM server via Docker. So you will need node running Docker. I will not convert the installation of docker in this tutorial.

Getting the PMM server up and running

Download the percona/pmm-server image

Login to your docker node and run

docker images 

output

root@docker2001:~# docker images
REPOSITORY          TAG                 IMAGE ID            CREATED             SIZE
nginx               latest              ae513a47849c        9 days ago          109MB
ubuntu              latest              0458a4468cbc        3 months ago        112MB

In my case I have only 2 images no percona/pmm-server image. Lets go ahead and download the percona/pmm-server image by issuing the command

 docker pull percona/pmm-server:latest

Now if we run again the commande "docker images" we should see the percona/pmm-server image

Output

root@docker2001:~# docker images
REPOSITORY           TAG                 IMAGE ID            CREATED             SIZE
nginx                latest              ae513a47849c        9 days ago          109MB
percona/pmm-server   latest              62f64eeb6eec        2 weeks ago         842MB
ubuntu               latest              0458a4468cbc        3 months ago        112MB

Start the percona/pmm-server container

Now that we have the image available we are going to create the pmm-data container and start it using the scripts below

Script to create the container

#!/bin/bash
docker create \
  -v /opt/prometheus/data \
  -v /opt/consul-data \
  -v /var/lib/mysql \
  -v /var/lib/grafana \
  --name pmm-data \
  percona/pmm-server:latest /bin/true

Script to start the container

#!/bin/bash
docker run -d \
  -p 80:80 \
  --volumes-from pmm-data \
  --name pmm-server \
  --restart always \
  percona/pmm-server:latest

Please refer to Docker documentation for the different options used in both scripts such as -v, -p and so on

make both scripts executable my running

chmod +x pmm-data.sh 
Chmod +x pmm-run.sh

Run both scripts

./pmm-data.sh
./pmm-run.sh

Now check the list of running containers

docker ps

Output

root@docker2001:~# docker ps
CONTAINER ID        IMAGE                COMMAND                CREATED             STATUS              PORTS                         NAMES
dfff2958b2b3        percona/pmm-server   "/opt/entrypoint.sh"   2 minutes ago       Up 2 minutes        0.0.0.0:80->80/tcp, 443/tcp   pmm-server

Open a browser and type in your IP address to access the pmm-server dashboard

Pmm-srv-dashboard.JPG

Install Pmm-client on each node

We are going to use the same nodes we used in How to install Percona XtraDB Cluster. We have a total of 3 nodes (db1,db2 and db3)

Process is the same, so I will be doing the first node (db1)

login to db1, run

sudo apt-get update 
sudo apt-get install pmm-client

Connect PMM client to PMM server

To connect each pmm client to PMM server we use the command "pmm-admin config --server IP_adress_of_server" in our case the command will be

pmm-admin config --server 10.192.0.96
OK, PMM server is alive.
PMM Server      | 10.192.0.96 
Client Name     | db1
Client Address  | 10.192.0.87 

10.192.0.96 being our PMM server IP addres

Note: if your PMM server has a username and a password the command to use is

pmm-admin config --server pmm_server_IP --server-user username --server-password password

After we connect the pmm client to the pmm server if you access the pmm server dashboard, you will not see Data from db1 or see db1 itself

We need to enable data collection from db1 so it can show up on the pmm sever.

Data collection

The metrics we need to see in the pmm server are: MYSQL metrics and MYSQL quuery analytics. To enable this, we run

pmm-admin add mysql --user root --password your_mysql_instance_password

Output

[linux:metrics] OK, now monitoring this system.
[mysql:metrics] OK, now monitoring MySQL metrics using DSN root:***@unix(/var/run/mysqld/mysqld.sock)
[mysql:queries] OK, now monitoring MySQL queries from slowlog using DSN root:***@unix(/var/run/mysqld/mysqld.sock)

Pmm-srv-db2.JPG

Pmm-srv-db3.JPG

After installing pmm client and enabling data collection on all 3 nodes we can now see all 3 node in the pmm server dashboard

Pmm-srv-db4.JPG

Query Analytics

Pmm-srv-db5.JPG

Troubleshooting

If you have n node in your cluster and you shutdown all n nodes by stopping first mysql (sudo systemctl stop mysql) and powering them off (sudo poweroff). When You you bring back all the nodes and try to start mysql, )sudo systemctl start mysql) mysql woundn't start. You need to bootstrap first the last node you stop mysql on. That will be the node that has safe_to_bootstrap: 1 in /var/lib/grastate.dat. Case: we have 3 nodes in our cluster in this tutorial (db1, db2 and db3) we stop mysql first on db1, db2 nd db3 . The node that will have it's safe_to_bootstrap: 1 will be db3 so we will have to run /etc/init.d/mysql bootstrap-pxc on db3 first and start mysql on db1 and db2 with sudo systemctl start mysql

/var/lib/mysql/grastate.dat
# GALERA saved state
version: 2.1
uuid:    56924696-1539-11e8-ab53-0ad44dbbac83
seqno:   -1
safe_to_bootstrap: 1

Conclusion

We have installed Percona Monitoring and Management server on docker and configured all 3 nodes in the cluster to connect to it. We also set what we needed to monitor from the cluster.