sfko 0.1.1

A utility for long running semi-random stress tests of Zenko

# SFKO

I am become death, destoyer of clusters.

```
usage: sfko [-h] [-c] [-w]

I am become death, destoyer of clusters.

optional arguments:
-h, --help show this help message and exit
-c, --controller Start sfko in controller mode
-w, --worker Start sfko in worker mode

If no options are provided sfko will start in standalone mode
```


# Testing model

SFKO implements a randomized testing strategy. Test configuration is loosely defined in `Scenarios` and specifics are chosen at random during runtime.

### Scenarios

`Scenarios` consist of a list of `tests` and `checks` to execute along with options defining required resources.

Simple `Scenario`

```yaml
- name: Write 0B
required:
buckets: 1
objects:
size: 0B
tests:
- put
checks:
- check-backend
```

`Scenarios` with all the bells and whistles

```
- name: Replicate 10M
required:
buckets:
- replication:
- *AWS
- *GCP
clouds:
- *AWS
objects:
count: 10000
size: 10M
tests:
- put-replication
checks:
- check-replication-mpu
```

#### Scenario options

```
- name: Example Scenario
required:
buckets: 1 # buckets can be a integer sepcify the number of buckets to create
buckets: # buckets can be a list with each element specify a bucket
- replication: True # replication can be set as a boolean
- replication: # or a list specify possible backend clouds
- *AWS
- *GCP
clouds: # clouds specifies a list of backends to be used for created buckets
- *AWS
```

### Tests and Checks

Functions that either drive the cluster or check behavior are split into two groups: `Tests` and `Checks`.


### Anatomy of a `Test` or `Check`

At its root a `Test` or `Check` is simply a function that returns `True/False` based on success.
```python
@register_test('put')
def put_objects(bucket, objs):
for obj, data in objs:
obj.upload_fileobj(data)
return True

```

```python
@register_check('check-backend')
def check_backend(bucket_conf, objs):
if that_op_worked():
return True
return False
```

New `Tests` and `Checks` can be registered with the decorators `register_test` and `register_check` respectively. When called each check is pass two objects: a `Bucket` and `ObjectProxy` instance. These describe the generated bucket and objects chosen for this test.

A `Bucket` instance has the following attributes:

```
name # Name of bucket
backend # Instance of Backend, describes chosen bucket backend
replication # Instance of Backend, describes chosen replication target
trasient # bool whether transient source is enabled
expiration # bool whether lifecycle expiration is enabled
versioning # bool whether versioning is enabled
clouds # List of cloud constants describing possible backend choices
client # A high level boto3 Bucket client
```
A `Backend` has the following attributes

```
name # Human freindly name for this backend
type # A constant describing this backends type
bucket # The cloudside bucket for this backend
```

A `ObjectProxy` has the following attributes

```
objects # An iterator yielding a boto3 Object, and a open file descriptor of content
raw # An iterator yielding a bucket name, key name, and file size
client # A low level boto3 client
resource # A high level boto3 resource
```