Use this information to prepare a cluster that contains
ESS for adding functions such as NFS, SMB, Object, GUI, and Performance
Monitoring.
Before adding protocols to a cluster that contains ESS, ensure
the following:
- The cluster containing ESS is active and online.
- RHEL 7.x is installed on nodes that are going to serve as protocol
nodes.
- RHEL 7.x base repository is set up on nodes that are going to
serve as protocol nodes.
- Use the ESS GUI or CLI to create a CES shared root file
system.
Note: The CES shared root file system must be at
least 4 GB in size and it must not be encrypted.
- On all nodes that are going to serve as protocol nodes,
download and extract the IBM Spectrum Scale™ protocols
standard or advanced packages. For information on extracting
packages, see Extracting the IBM Spectrum Scale software on Linux nodes.
- On all nodes that are going to serve as protocol nodes, install core GPFS™ RPMs from the /usr/lpp/mmfs/5.x.x.x/gpfs_rpms.
Attention: Ensure that you do not install packages such as perfmon,
gui, callhome, java, and
protocols at this time. These components will be installed and configured with
the protocols deployment.
The core GPFS RPMs that need to be installed include:
- gpfs.base
- gpfs.ext
- gpfs.gpl
- gpfs.license.xx
- gpfs.gskit
- gpfs.docs
- gpfs.msg.en_US
- gpfs.compression
- gpfs.adv (optional)
- gpfs.crypto (optional)
- Add the nodes that are going to serve as protocol nodes
to the cluster using the mmaddnode command.
- Enable the Cluster Configuration Repository (CCR) on the
cluster, if it is not enabled already.
Attention: If GPFS levels between protocol nodes and ESS differ significantly, ensure that the nodes with the newer code level of GPFS are designated as both quorum and manager nodes. For example, old ESS systems with GPFS 4.1.0-8 are incompatible with CES. These ESS systems can be a part of a cluster with protocols only if they are not designated as quorum and manager nodes.
This
aspect requires careful planning and administration because:
- By default, ESS nodes are designated as quorum and manager nodes.
- In some cases, there might not be extra nodes outside of ESS to
make sure that only nodes with the newer code level are designated
as quorum and manager nodes.
- Verify that passwordless SSH is working between all nodes in the cluster.
- Verify that firewall ports are set correctly.
- Verify that the CES shared root file system is mounted
and that it is set to auto mount.
- Set NFSv4 ACLs for file systems that are going to be used
for protocol data, if they are not set already.
- On one of the nodes that are going to serve as protocol
nodes, run the installation toolkit to
start deploying protocols.
Note: - Designate only the nodes that you have planned to use as protocol
nodes. Do not designate existing ESS nodes such as EMS or I/O nodes
as protocol nodes.
- Point to existing file systems only. Do not attempt to create
new file systems or NSDs for this purpose.
For information about deploying
protocols in an existing cluster using the installation toolkit, see Deploying protocols on an existing cluster and spectrumscale command.