My work based on enhancing and fixing issues with https://github.com/IBM/StorageScaleVagrant/
This uses a freely downloadable version of IBM Storage Scale You can donwload a copy (after filling in a short form) from https://www.ibm.com/account/reg/us-en/signup?formid=urx-41728
These are some of the key objectives
1 Automate the install of Storage ScaleVagrant including the: git download, basbox build, and build of the Storage Scale server Also I want to skip the automated running of the built-in demos so I can either:
x Run each one on demand, possibly menu driven (Curses anyone?)
x Run my own demos customised for the client's interests
2 Speed up the deployment time to be able to demonstate features as soon as possible by:
x Have a reliable Basebox stored bentrally that I can build both servers and clients from.
x Avoid repeating slow steps such as `spectrumscale {install, deploy} eg ny detecting that they have already been run
x Enable to GUI as soon as it is deployed
x (todo) create a Boxfile of the server VM so can spin up say the S3 demo very quickly. Also consider the use of VirtualBox snapshots here?
3 Try and minimise any modifcation of StorageScaleVagrant
. This is beause I don't want to risk incompatability as new drops of this come out every few months. Currently the main think I change is to disable the automated running of the built-in demos.
4 Be able to demonstate as many data access protocols as possible. Also remote management so far we have:
x POSIX using native GPFS via Mulicluster
x S3 using the AWS toolkiy and he new Nooba based S3 sever in Storage Scale CES
x NFS : as standard client server
x (Todo) HDFS - although this may be difficult client side
x https RestAPI for management
x (Todo) New role based access to the 'mm' commands
The only prerequisites are freely downloadable copies of
- VirtualBox
- Vagrant
- Spectrum Scale Developer edition tarball
Firstly create the basbox which contains a copy of Centos/8 with added to the RPMs that Storage Scale depends on. We also install some nice to have RPMs like unzip
and jq
./install_basebox.sh
Then we can install in either order the server
./install_scale.sh
and the client
./install_client.sh
Then the following can be run in any order:
./setup_multicluster.sh
./setup_mms3.sh
./setup_nfs.sh
The original demo suite ''' ./run_demos.sh '''
My work (in progress) on teh Scale RestAPI
./install_restapi
./test_restapi.ipynv
or directly
./test_restapi.sh
Vagrant ssh
is quite slow as it loads a lot of stuff. It is better and far quicker to run ssh directly.
But to do this we need to remember where we stored the ssh keys for passwordless login.
Also as the VM can get re-installed regularly we don't want erroneous security warnings about the host's SSL keys having changing
This is a great gem - much faster that vagrant ssh
alias fastssh="ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null -i .vagrant/machines/*/virtualbox/private_key -p 2200 [email protected]"
Actually the methid I prfer much more than this is to populate ~/.ssh/config on my laptop (Git Bash or Mobaxterm).
Here we will ssh direct to the VM's IP address rather than a forwardd port like 20222 on localhost