Showing results for 
Show  only  | Search instead for 
Did you mean: 

HTTP monitors: how is TCP connect time calculated when two connections exist?


In Custom Charts, it is possible to graph the connection time of an HTTP monitor, by using the following metric:

HTTP monitor - TCP connect time (by location) (Average)

In my case, the HTTP monitor does two TCP connections:

  1. First one is to the initial server, which then does a redirect to a different host
  2. Second one, where the HTML is produced

I'm not sure which value the custom metric represents, but given the latencies involved, I have made the following reasoning:

  1. As I have a very small latency for connection #1, and TCP connect time is high, it cannot be the time for the first connection
  2. Might be the time for the second connection
  3. or, it might be the sum for both connections

As I couldn't find it in the documentation, does someone know if it's #2 or #3?



There are two metrics for the HTTP monitors for TCP connection. One is general per HTTP monitor and the second one is for request.

Based on your question, I believe you selected the first one (I assume it's average of the steps in a monitor) and you probably want the second one (TCP connection time per request).

Nice! Hadn't noticed it. But the values are the same.

I believe the one by request/step, is for HTTP monitors that have more than one request... That is not my case, as redirects are counted as only one request.