We are currently experiencing highly frequent timeouts from some of our machines. The symptoms are rather identical to what is described in issue 100354. I.E. our NodeJS app emits ETIMEDOUT when trying to connect to the api.twitter.com IP addresses.
curl https://api.twitter.com from the same machines that fail, we are seeing a 100% success rate in receiving response. So this might be a problem with some specific endpoints only, or with NodeJS apps only.
Unfortunately, I cannot provide ping/traceroute/mtr logs due to ICMP limitations on the network we cannot circumvent at the time being.
TCP dumps show that a connection is actually established with the twitter API endpoint, and that data is being transmitted. But since it is encrypted I cannot tell what the data contains. The session is finally terminated with FIN, ACK packages which are then followed by RST packages.
Since curl works, and since we are able to flawlessly connect from other machines on other networks, our immediate suspicion is that the problem is not on Twitter’s end. Nonetheless, I am posting here to see if anybody else encountered similar problems and had any luck resolving them. Also, if the Twitter staff can assist us with debugging (such as checking what happens to the sessions on your end) this it would be tremendous help!