table of contents Table of contents
search
Checkly docs / API checks / Timeouts timing

Timeouts and timing phases

On this page

All API checks are capped at a timeout of 30 seconds. With each request, we record the most relevant timing phases. This can help you troubleshoot slow responses, e.g. your DNS might be slow.

The timing phases correspond to the Node.js request library timing phases:

  • wait: Duration of socket initialization
  • dns: Duration of DNS lookup
  • tcp: Duration of TCP connection
  • firstByte: Duration of HTTP server response
  • download: Duration of HTTP download

api monitoring timing phases

“wait” is the time it takes for the underlying Node HTTP module to get a socket from the network layer. This time can vary quite a bit, usually it goes from 900 microseconds to some milliseconds. Most clients for API / webpage will have this lag too.

Error: ESOCKETTIMEDOUT. Sometimes API checks return this error, without any other information on what caused it.

  • This is a socket timeout error. Essentially there is no successful connection with the API at the TCP and/or DNS level. Then, Checkly is closing the connection after the usual 30 seconds timeout, because the server didn’t respond.
  • Usually the root cause for this error is intermittent network issues on the application side.
  • There will probably be no errors in the application logs, as it was not even possible to establish a connection.
  • The response headers will be empty. If Checkly doesn’t get a connection, there is nothing to send back, e.g. the response headers.
Setup & teardown scripts

You can contribute to this documentation by editing this page on Github