With a lot of concurrent requests to metadata service, credentials can fail

[mick]

This is encountered as CredentialsError: Missing credentials in config because the request for IAM Role can timeout before all the requests are answered. fixes #445

The fix:

• add a locking so the metadata service isnt flooded with requests.
• add setImmediate so IO can still be processed even with a lot of request to the metadata service being generated.

This is a bit hard to add a test for. I did add a test for multiple concurrent requests to loadCredentials.

On an ec2 of the same type and size: This test case passed with these changes, even with increasing concurrency up to 10000.

cc: @lsegal @willwhite

[coveralls]

Coverage increased (+0.01%) when pulling 2698bba on mick:creds_lock into 8a15f08 on aws:master.

[coveralls]

Coverage increased (+0.01%) when pulling 0c70713 on mick:creds_lock into b768e76 on aws:master.

[mick]

@lsegal I think this is all set to go, unless there is anything else you think is blocking. Thanks!

[lsegal]

Looks good, thanks!

[mick]

@lsegal great! Just curious, what does the release / publish cycle for aws-sdk look like? I'd like to be able to use this from npm.

[lsegal]

We have a release scheduled this week. FWIW you can use this from npm already if you npm install git://github.com/aws/aws-sdk-js.

[paritosh16]

I have a number of node modules that are running as cron jobs. The modules use the aws-sdk for javascript (Installed using npm install aws-sdk, current version being 2.1.38). It is giving the following error randomly,

{ [TimeoutError: Missing credentials in config]
  message: 'Missing credentials in config',
  code: 'CredentialsError',
  time: Mon Jul 13 2015 09:20:02 GMT+0000 (UTC),
  originalError:
   { message: 'Could not load credentials from any providers',
     code: 'CredentialsError',
     time: Mon Jul 13 2015 09:20:02 GMT+0000 (UTC),
     originalError:
      { message: 'Connection timed out after 1000ms',
        code: 'TimeoutError',
        time: Mon Jul 13 2015 09:20:02 GMT+0000 (UTC) } } }

This occurs randomly for all the modules. After the error, it again works fine for some time with another occurence of the error. Please help.

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs and link to relevant comments in this thread.