Add ability to shorten check confirmations
It would be great if you could have a check that checks every 10 minutes, but when it starts failing you can have do the confirmation checks every minute.
At the moment (I could be wrong here) if I have it set to check every 10 minutes it would take 30 minutes of downtime for me to get a notification.
Actually the confirmation checks are twice as fast, which means that with a 10 minute period, you’ll be alerted 10 to 20min after the downtime started (and only for downtimes longer than 10 min)
If you check every 10m it means a downtime up to 10m is acceptable. Running confirmation checks faster would mean being sensitive to downtimes smaller than 10min.
If you want faster alerts, you need to select a faster check interval, there’s no point in being notified in 2 min if you check every 10 minutes anyway because If you care about a 2 min downtime at 12:00, you should also care about a 2 min downtime at 12:05 too ;)
In the end, we chose to keep a precise balance between sensitivity and frequency so you can’t be notified for a downtime randomly, this is important because with a 10 minutes interval you can miss a downtime below 10 min, so even if we catch them we must keep the same sensitivity so the alerts aren’t random.
I hope you understand this is engineered in your best interest, for the monitoring to be reliable and consistent.
-
Sure, definitely a good idea, but in that case you will get alerted in 10 to 20 min.
-
James commented
Thanks for the reply! I agree with your reasoning and it makes sense. The only reason I'm currently using 10 minute intervals is for checks on platforms like Cloudfront/S3 where you pay per request, so keeping it low keeps costs low.