This is a script which parses the 3rd party spotinfo,
as well as AWS data from the describe instance-types API.
It then produces a table, CSV output or raw JSON to help you work out what instance types to specify when setting up an autoscaling group.
By default, the hardware architecture x86_64 is selected, and the hypervisor nitro is also
selected. Instances which are not eligable for spot are hidden, as are instances which don't have
interruption data (from spotinfo). There is also a minimum spot interruption threshold of 10%.
To see how to adjust these values, run ./spot-instances-review.py --help.
You are recommended to use a virtual environment, so first run
virtualenv .venv && source .venv/bin/activate. On subsequent runs, you will only need to run
source .venv/bin/activate before running the script.
This script uses Python3 and a couple of python libraries, so ensure you have python3 and pip
installed, and then run pip install -r requirements.txt.
Running ./spot-instances-finder.py will use the following criteria:
INFO:root:Search criteria:
INFO:root:Memory : >0.1 <10000.0
INFO:root:CPU : >0.1 <10000.0
INFO:root:architecture: x86_64
INFO:root:hypervisor : nitro
INFO:root:interruption: <10
You can use additional flags to change these values. As a shorthand, you can also use the flag
./spot-instances-finder.py --like t3.2xlarge which will show you all the options which are
directly comparable to the t3.2xlarge instance size (for example).
If you run ./spot-instances-finder.py --order-by-price this will order the results by the
spot price. Run it ./spot-instances-finder.py --order-by-interruption and it will return the
results by the risk of interruption.