Skip to content

Commit 09ff5ce

Browse files
Update README.md
1 parent fa62311 commit 09ff5ce

File tree

1 file changed

+18
-14
lines changed

1 file changed

+18
-14
lines changed

README.md

Lines changed: 18 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -24,42 +24,46 @@ Salt configuration is stored under ./salt directory, containing pillar/ (variabl
2424
![Architecture](images/architecture.png)
2525

2626
### Operations
27-
2827
Salt commands should be executed from the headnode.
2928
IntelMPI installation: sudo salt '*' state.apply intelmpi
3029

31-
3230
### SSH Key
33-
3431
SSH key is generated each time for the environment in the ./key.pem file.
3532

36-
3733
### Networking
3834

3935
- Public subnet
40-
4136
Headnode acts a jump host and it's placed in the public subnet. The subnet is open to SSH connections from everywhere. Other ports are closed and can be opened using custom-security-list in the OCI console/cli.
4237
All connections from VCN are accepted. Host firewall service is disabled by default.
4338

4439
- Private subnet
45-
4640
All connections from VCN are accepted. Public IP's are prohibited in the subnet. Internet access is provided by NAT gateway.
41+
42+
### Roles
43+
44+
- intelmpi: provides configured Intel yum repository and installs IntelMPI distribution
45+
- openmpi: installs OpenMPI from OL repository
4746

4847
### Storage
48+
- Storage node require to be DenseIO shape (NVME devices are detected and configured).
4949

50-
- Block volumes
50+
#### Filesystems
5151

52+
- GlusterFS (requires storage role)
53+
To use BeeGFS set storate_type to beegfs.
54+
Filesystem will be greated as :/gfs and mounted under /mnt/gluster
55+
56+
- BeeGFS (requires storage role)
57+
To use BeeGFS set storate_type to beegfs.
58+
59+
#### NFS
60+
61+
- Block volumes
5262
Each node type can be configured with block volumes in the variables.tf
5363
Headnode will export first block volume as NFS share under /mnt/share (configured in salt/salt/nfs.sls)
5464
Other block volume attachments need to be configured manually after cluster provisioning.
5565

56-
- GlusterFS
57-
58-
To use Gluster specify Gluster shape and node count in the variables.tf
59-
Filesystem will be greated as :/gfs and mounted under /mnt/gluster
60-
6166
- FSS
62-
6367
File system service endpoint will be created in the private subnet and mounted on each node under /mnt/fss
6468

65-
69+

0 commit comments

Comments
 (0)