Symptom
Catchpoint allows users to pass/request and also capture HTTP header information through tests. If Capture HTTP Headers is enabled, the header information captured from the waterfall page is accessible by clicking on the URL in the record.
There are specific scenarios where, in a Transaction test, the ‘Headers’ page shows up empty and does not display any header information even though the selection was made to capture it.
Cause
This can occur in cases where the test was retried and then failed. When a retry occurs, if the upload to the capture server fails, the capture server will failover to choose the first capture data for the given test, node, and schedule window, even if the actual runtime doesn't match the given capture server data. This results in the display of a blank page.
Example
Here's a typical scenario where a test may be retried, resulting in missing header information:
A test takes multiple steps to run (e.g. 13 steps, with a ~60 second average to run), and the verify-on-error switch is turned on. If an error occurs, the test can easily take two minutes to finish running. When a test takes more than 2.5 minutes to complete, it may get retried.
There are other scenarios where the test may get retried:
- If the test takes too long to respond to the schedule server due to node-specific network issues, node-specific heavy load issues, or test-specific network issues.
- If the node machine or node service restarts.
- If the schedule server restarts, then all tests can possibly re-run shortly after the restart.
- If there is a cutover from one schedule server to a backup schedule-server (via DNS), it may cause tests to re-run.