Whenever a site encounters an error condition while taking a sample — for example, a search string is not found on a page, or a page takes longer than the given timeout value to load — then that sample could be logged as an error or strike and may cause an alert to be generated.
By default, our system uses a three-strike rule to determine whether to send an alert, which prevents extraneous alerts due to intermittent or localized problems. The three-strike rule works as follows:
You can configure the number of strikes before error from 1 to 4, depending on your alerting needs. To change the Strikes Before Error setting:
If you have changed the number of strikes before error, then an alert will be sent on the final failed sample. For example, if the number of strikes before error is set to 1, then you will receive an alert every time an error condition is encountered; if the strikes before error is set to 4, then you will receive an alert on the fourth failed attempt to gather a sample. The final failed sample will be recorded in the service's log files as an error.
If you are monitoring simultaneously, instead of sequentially as described above, samples are taken all at once from different locations and an alert is issued if the appropriate number of the samples taken are unsuccessful, based on the number of strikes.
(Baseline and Additional Agents are configurable in the Services tab > Settings sub tab > Monitoring Agents list.)