support --log / -l for invoke local (especially for Python logging)

[ceefour]

Feature Proposal

Description

Currently Python logging doesn't output anything using invoke local

Add logging support similar to invoke -l

Especially with Python logging

I'd argue that --log for invoke local should be enabled by default, rather than needing to be enabled explicitly.

Additional Data

• Serverless Framework Version you're using: 1.20.2
• Operating System: Windows 10

[dschep]

I don't quite follow what you're asking for. sls invoke local already shows stderr and stdout, unlike sls invoke so there's no need to add a --log option AFAICT.

Here's the aws-python3 template with print('foo') added:

$ sls invoke local -f hello
foo
{
    "statusCode": 200,
    "body": "{\"message\": \"Go Serverless v1.0! Your function executed successfully!\", \"input\": {}}"
}
$ sls invoke -f hello
{
    "statusCode": 200,
    "body": "{\"message\": \"Go Serverless v1.0! Your function executed successfully!\", \"input\": {}}"
}
$ sls invoke -l -f hello
{
    "statusCode": 200,
    "body": "{\"message\": \"Go Serverless v1.0! Your function executed successfully!\", \"input\": {}}"
}
--------------------------------------------------------------------
START RequestId: 1e2d9cd9-8e90-11e7-bf98-a994067d9d91 Version: $LATEST
foo
END RequestId: 1e2d9cd9-8e90-11e7-bf98-a994067d9d91
REPORT RequestId: 1e2d9cd9-8e90-11e7-bf98-a994067d9d91  Duration: 0.39 ms       Billed Duration: 100 ms    Memory
 Size: 1024 MB    Max Memory Used: 21 MB

and using the python logging module:

[:wat] $ sls invoke local -f hello
ERROR:root:foo

{
    "statusCode": 200,
    "body": "{\"message\": \"Go Serverless v1.0! Your function executed successfully!\", \"input\": {}}"
}
$ sls invoke -f hello
{
    "statusCode": 200,
    "body": "{\"message\": \"Go Serverless v1.0! Your function executed successfully!\", \"input\": {}}"
}
$ sls invoke -l -f hello
{
    "statusCode": 200,
    "body": "{\"message\": \"Go Serverless v1.0! Your function executed successfully!\", \"input\": {}}"
}
--------------------------------------------------------------------
START RequestId: aa1afece-8e90-11e7-a0a7-fbe2e186fe49 Version: $LATEST
2017-08-31 17:09:23.822 (-04:00)        aa1afece-8e90-11e7-a0a7-fbe2e186fe49    [ERROR] foo
END RequestId: aa1afece-8e90-11e7-a0a7-fbe2e186fe49
REPORT RequestId: aa1afece-8e90-11e7-a0a7-fbe2e186fe49  Duration: 1.02 ms       Billed Duration: 100 ms    
emory Size: 1024 MB     Max Memory Used: 21 MB

[ceefour]

@dschep mine behavior is different:

def search(event, context):
    logger = logging.getLogger('posts')
    logger.setLevel(logging.DEBUG)
    logger.info('log INFO event: %s', event)
    logger.error('log ERROR event: %s', event)
    print('print event: %s' % event)

Result:

C:\Users\ceefour\git\modelooks\serverless>sls invoke local -s local -f postsSearch
log ERROR event: {}

print event: {}

[dschep]

@ceefour, that's odd. I'm not sure why it'd be doing that 😕

[ceefour]

@dschep there is light:

def search(event, context):
    logger = logging.getLogger('posts')
    logger.setLevel(logging.DEBUG)
    ch = logging.StreamHandler()
    ch.setLevel(logging.DEBUG)
    logger.addHandler(ch)
    logger.info('log INFO event: %s', event)
    logger.error('log ERROR event: %s', event)
    print('print event: %s' % event)

gives:

C:\Users\ceefour\git\modelooks\serverless>sls invoke local -s local -f postsSearch
log INFO event: {}

log ERROR event: {}

print event: {}

which is kinda acceptable if I'm so inclined. But I don't get why I have to do so much boilerplate while you (and AWS) can get sane nice defaults ?

Perhaps this is not so much request for -l but rather better support for ... Python on Windows (?) maybe.

[dschep]

Hmm. I'm guessing AWS does something in their python code before it imports and calls your handler, invoke.py could probably do the same/similar if someone knows what setup AWS is doing.