azurehpc/examples/beegfs_pools
Xavier Pillons 0c9e9ca6d3 add cndefault on all storage nodes 2020-03-16 17:35:51 +01:00
..
config.json add cndefault on all storage nodes 2020-03-16 17:35:51 +01:00
readme.md Removed azhpc-view and put links for azurehpc-ui 2019-09-13 15:58:39 +01:00

readme.md

BeeGFS Cluster built using BeeGFS pools, ephemeral disks (temporary) and HDD disks (persistent)

Visualisation: config.json

This will deploy a BeeGFS PFS using ephemeral disks (L8s_v2) and the attached Standard HDD disks, BeeGFS pools will be set-up to enable moving data between the ephemeral disk and the HDD permanent disk. A headnode, an NFS server running on the headnode (User accounts shared home directories will be stored here), compute cluster and PBS will be also be deployed.

NOTE:

  • MAKE SURE YOU HAVE FOLLOWED THE STEPS IN prerequisite before proceeding here

First initialise a new project. AZHPC provides the azhpc-init command that will help here. Running with the -s parameter will show all the variables that need to be set, e.g.

$ azhpc-init -c $azhpc_dir/examples/beegfs_pools -d beegfs_pools -s
Fri Jun 28 08:50:25 UTC 2019 : variables to set: "-v location=,resource_group="

The variables can be set with the -v option where variables are comma separated. The -d option is required and will create a new directory name for you.

azhpc-init -c $azhpc_dir/examples/beegfs_pools -d beegfs_pools -v location=westus2,resource_group=azhpc-cluster

Create the cluster

cd beegfs_pools
azhpc-build

Allow ~15 minutes for deployment.

To check the status of the VMs run

azhpc-status

Connect to the Beegfs master node (beegfsm) and check that the BeeGFS pools were set-up

$ azhpc-connect -u hpcuser beegfsm

$ beegfs-ctl --liststoragepools
Pool ID   Pool Description                      Targets                 Buddy Groups
======= ================== ============================ ============================
      1            Default 2,4
      2           hdd_pool 1,3

Writing/reading to /beegfs/hdd_pools will use the HDD disks and if you use /beegfs the ephemeral disks are used.

$ beegfs-df
METADATA SERVERS:
TargetID   Cap. Pool        Total         Free    %      ITotal       IFree    %
========   =========        =====         ====    =      ======       =====    =
       1      normal    1787.6GiB    1787.6GiB 100%      178.8M      178.8M 100%
       2      normal    1787.6GiB    1787.6GiB 100%      178.8M      178.8M 100%

STORAGE TARGETS:
TargetID   Cap. Pool        Total         Free    %      ITotal       IFree    %
========   =========        =====         ====    =      ======       =====    =
       1      normal    4093.7GiB    4093.7GiB 100%      409.6M      409.6M 100%
       2      normal    1787.6GiB    1787.6GiB 100%      178.8M      178.8M 100%
       3      normal    4093.7GiB    4093.7GiB 100%      409.6M      409.6M 100%
       4      normal    1787.6GiB    1787.6GiB 100%      178.8M      178.8M 100%

Connect to the headnode and check PBS and BeeGFS (it will be mounted at /beegfs)

$ azhpc-connect -u hpcuser headnode

Resources:

  • Head node (headnode)
  • Compute nodes (compute)
  • BeeGFS
    • Management server (beegfsm)
    • Object storage servers and metadata servers(beegfssm)

Note: The Hc nodes are used for the cluster. To get best performance nodes with accelerated networking should be used.