install-cluster.md

Steps for installing a new cluster

Installing gluster on the servers

• Make sure you can ping the new machines to be installed:

  ./ansible-ec2 -l cluster1,cluster2 ping.yml

• Register the servers w/ subscription manager and update (Warning: this will reboot all servers in the list):

  ./ansible-ec2 -l cluster1,cluster2 register-servers.yml

• Install the servers.
  While it is possible to do this by directly running site.yml, it is recommended that the sub-playbooks be run individually to more easily supervise the process. (current list follows)

  ./ansible-ec2 -l cluster1,cluster2 playbooks/ssh-keys.yml
  ./ansible-ec2 -l cluster1,cluster2 playbooks/install-servers.yml
  ./ansible-ec2 -l cluster1,cluster2 playbooks/gluster-volume.yml
  ./ansible-ec2 -l cluster1,cluster2,pcp-aggregators playbooks/install-jumphost.yml

Creating the subvol PVs

After the supervols are online (from the above procedure), create the subvol PVs:

• Run the ansible playbook to create the PVs:

  # To create PVs for all supervols on the cluster:
  $ ./ansible-ec2 -l cluster1 playbooks/create-subvol.yml
  # ... or ...
  # To create PVs for just one supervol:
  $ ./ansible-ec2 -l cluster1 -e only_volumes=myvol playbooks/create-subvol.yml

• Copy the yaml file that contains the PV description to your local machine. You will need to use scp through the jump host. Assuming the jump host is 1.2.3.4 and the server we are working with is 192.168.20.11, from your local machine:

  $ scp -o ProxyCommand="ssh -W %h:%p -q [email protected]" [email protected]:/mnt/supervole1b00/pvs-0-4999.yml .

Creating the subvol PVs (old)

The below procedure has been superceeded by the ansible playbook above

• Log into the 'master' machine for the cluster

• Start screen (this will take a while to run):

  $ screen -xR

• Check out the subvol code:

  $ git clone https://github.com/gluster/gluster-subvol.git

• Mount the volume on the host:

  $ sudo gluster vol list
  gluster_shared_storage
  supervole1b00
  $ sudo mkdir /mnt/supervole1b00
  $ sudo mount -tglusterfs localhost:/supervole1b00 /mnt/supervole1b00

• Run the command for the creator script. The proper syntax can be found by looking at the top of the creator_xfs_quota.sh script. Below, we are creating PVs 0 - 4999, each w/ a 1 GB quota.

  $ cd /mnt/supervole1b00
  $ sudo /home/ec2-user/gluster-subvol/volcreator/creator_xfs_quota.sh 172.31.80.251:172.31.87.134:172.31.93.163 supervole1b00 /mnt/supervole1b00/ 1 0 4999

• The above command will create 2 files:

  • pvs-0-4999.yml, which is the PV descriptions
  • quota-0-4999.dat, which is the description of the quotas to apply to the bricks

• On each brick server, apply the quotas by providing the brick directory and the quota file to the apply_xfs_quota.sh script

  $ sudo /home/ec2-user/gluster-subvol/volcreator/apply_xfs_quota.sh /bricks/supervole1b00/brick /mnt/supervole1b00/quota-0-4999.dat

• Once the command completes, verify that all quotas have been properly created.

  $ sudo xfs_quota -x -c 'report -p -a'

• Copy the yaml file that contains the PV description to your local machine. You will need to use scp through the jump host. Assuming the jump host is 1.2.3.4 and the server we are working with is 192.168.20.11, from your local machine:

  $ scp -o ProxyCommand="ssh -W %h:%p -q [email protected]" [email protected]:/mnt/supervole1b00/pvs-0-4999.yml .

Next steps

• Package up the PV descriptions, recycler config, and TLS keys. These need to be sent to the Online team so that the OpenShift side can be deployed.

• Make sure the newly provisioned hosts are reporting into Zabbix properly

• Check back in a couple days and make sure snapshots are working properly.

  $ sudo gluster snap list