OCI Available Capacity Tools

This repo contains tools to help display OCI available capacity in a tenancy and randomly choose available shapes.

Prerequisites

• Install Python 2.7 or later.
• Install required Python packages:

pip install -r requirements.txt

• Details (OCID, private key, fingerprint) for a user in the "Auditors" or "Administrators" OCI group.

Config File

Create a config file containing tenancy and user details in the "OCI" section, for example:

[OCI]
user: ocid1.user.oc1..aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
tenancy: ocid1.tenancy.oc1..aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
key_file: /tmp/my_api_key.pem
fingerprint: aa:aa:aa:aa:aa:aa:aa:aa:aa:aa:aa:aa:aa:aa:aa:aa

And sections for the service limits of each region: a list of compute shapes, and their capacities per-AD, for example:

[us-phoenix-1]
VM.Standard1.1: 5
VM.Standard1.2: 5
VM.Standard1.4: 5
VM.Standard1.8: 5

[us-ashburn-1]
VM.Standard1.1: 3
VM.Standard1.2: 2
VM.Standard1.4: 5
VM.Standard1.8: 1

Currently, these limits must be manually retrieved from the OCI Console: Region->Manage Regions->Service Limits. Once OCI exposes a REST API to retrieve these, this tool can be changed to use that, thus eliminating the need for the above sections of the config file.

Available Compute Capacity Tool

This tool prints out the currently available compute capacity for each shape:

python compute_capacity.py --config /tmp/my_config

The output is something like this, showing the available capacity for each region/shape/AD:

************************************************
Available shapes in us-ashburn-1
************************************************
VM.Standard1.1: AD1: 3, AD2: 1, AD3: 1
VM.Standard1.2: AD1: 0, AD2: 1, AD3: 1
VM.Standard1.4: AD1: 5, AD2: 1, AD3: 4
VM.Standard1.8: AD1: 0, AD2: 0, AD3: 1
************************************************
Available shapes in us-phoenix-1
************************************************
VM.Standard1.1: AD1: 3, AD2: 1, AD3: 1
VM.Standard1.2: AD1: 0, AD2: 1, AD3: 2
VM.Standard1.4: AD1: 4, AD2: 2, AD3: 2
VM.Standard1.8: AD1: 0, AD2: 1, AD3: 1

Compute Randomizer Tool

This tool randomly picks a region and set of shapes/ADs that satisfy a set of compute "requests". A compute request is defined as a logical name, and a list of compute instances required in separate ADs. The user-specified "requests" is then of the following form: "<request1_name>:<comma_separated_instances_per_ad> <request2_name:<comma_separated_instances_per_ad> ...". Any region able to satify all requests has equal chance of being chosen. And any shape/ad_list within the region that is able to satisfy a given request has equal chance of being chosen, with the most demanding requests satisfied first.

A couple of examples for "requests":

• --requests "webtier:2" will choose a shape for "webtier" with 2 instances in one AD.
• --requests "webtier:2,1,3" will choose a shape for "webtier" with 2 instances in one AD, 1 instance in another, and 3 in another.
• --requests "master:1,2 worker:2,1,3" will chose:
  • A shape for "master" with 1 instance in one AD and 2 instances in another.
  • A shape for "worker" with 2 instances in one AD, 1 instance in another, and 3 in yet another.

python compute_randomizer.py --config /tmp/my_config --requests "master:1,2 worker:2,1,3"

The output of the tool shows the chosen region, and for each request, the chosen shape and AD list that satisfy the request. For example:

REGION=us-ashburn-1
MASTER_SHAPE=VM.Standard1.4
MASTER_AD1=1
MASTER_AD2=3
MASTER_ADS=1,3
WORKER_SHAPE=VM.Standard1.8
WORKER_AD1=3
WORKER_AD2=1
MASTER_AD3=2
MASTER_ADS=3,1,2