You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@usergrid.apache.org by "ryan bridges (JIRA)" <ji...@apache.org> on 2015/01/27 21:06:34 UTC

[jira] [Created] (USERGRID-368) A connection timeout to the configured mail server causes POST requests to /management/organizations to block

ryan bridges created USERGRID-368:
-------------------------------------

             Summary: A connection timeout to the configured mail server causes POST requests to /management/organizations to block
                 Key: USERGRID-368
                 URL: https://issues.apache.org/jira/browse/USERGRID-368
             Project: Usergrid
          Issue Type: Story
            Reporter: ryan bridges


A connection timeout to the configured mail server causes POST requests to /management/organizations to block. This can cause downstream load-balancers, proxies, and clients to also timeout even though the call eventually completes successfully.

In the case of a load-balancer, this causes even more confusion. Consider the following scenario:
- client makes a post request to the /management/organizations endpoint on the load-balancer
- the load-balancer makes the request to Usergrid instance #1
- Usergrid #1 successfully creates the org, app, and admin user, then attempts to contact the unavailable SMTP server
- After 15s, the load-balancer cancels the initial request and makes and identical request to Usergrid #2
- Usergrid #2 sees that the requested organization already exists and returns a 400 duplicate indexed property error to the load-balancer
- the load-balancer returns the 400 error to the client

The originating client application would only see the 400 error, even though the request completed successfully.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)