Update Wednesday, 11:17 PM EST: Fingers crossed. It looks as if the Twitter API issue has been solved.
Around last week Wednesday (March 25) we started noticing an increasing number of Twitter API calls getting an HTTP response code of 0 from the PHP cURL call to the API.
On Friday (March 27) the issue reached such proportions that we submitted a Twitter API issue. It was clear at that point that it wasn’t just a temporary glitch in the matrix.
Initially, the Twitter folks thought the issue was a networking problem on our side of the fence, which is quite understandable.
We, with the help of our hosting provider’s technicians, ran tests to troubleshoot the issue.
On Saturday morning we ran conclusive tests that showed the following:
- Calls to the Twitter Search API and to any other web site went through without any issue.
- Calls to the Twitter API went through sometimes, and other times received a Connection Refused, which resulted in the HTTP response code of 0 in the PHP cURL call.
Those tests showed that it was not a networking issue on our side, because only calls to the Twitter API experienced the problem.
Since it was weekend, not much further could be done at that point. We sent through the details to Twitter.
On Monday Twitter further investigated and determined that the issue was not with the site rate limit of TweetLater. We were well within the 20,000 API calls per hour limit.
The logical conclusion was that something somewhere on the Twitter network is (seemingly) randomly blocking API calls from our IP address. This could be a switch, router, firewall, etc.
Currently the Twitter operations staff and hosting provider technicians are investigating to see if they can pinpoint the problem.
There is unfortunately no time to resolution yet, until such time that they can actually pinpoint the root cause of the problem.
We’ll update this post with new information as it becomes available.
We sincerely apologize for the inconvenience, but please understand that the problem is out of our hands.