Diagnostics Mediator - Server Request Rate Watchdog Settings
Does anyone see any value in the Mediator alerts for "Excessive number if Server Requests", which are generated based on the Mediator server.properties "Server Request Rate Watchdog Settings "?
The default values are showne below - some of these seem arbitrary. Are there any rules or other recommended methods for calculating different thresholds? Using the defaults, i receive alerts on a daily basis, but the Mediators seem to be performming well and no issues encountered performance-wise anywhere else.
# Server Request Rate Watchdog Settings
# How often to check for unique server requests. Default is 1d. Set to 0 to disable.
watchdog.server_request_rate.frequency = 1d
# If the number of SRs in the 1M crosses this threshold an email is sent to the admin
watchdog.server_request_rate.threshold.1M = 400
# If we have more than 5x the number of thresholds in the 1M bucket vs the 7d bucket an email is sent to the admin
watchdog.server_request_rate.threshold.multiplier.7d.1M = 5
# If we detect more than 100 rest type SRs (e.g. SRs with a constant length or a number at the end) an email is sent to the admin
watchdog.server_request_rate.threshold.threshold_restUrls.7d = 100
The idea is to control and alert when you are recieving too many unique server request since each of them would be a symbol on the Time Series Database. You are correct, this would depend on your server too (if you dont have too many probes connected then your server would be able to handle more per probe).
You can run the following URL to get a count per probe.
Each probe (java or .net) has a URL collapsing in case there is static content or URLs with just a customer ID difference or any other criteria that can cause multiple unique URLs that are the same at the end. So this alert can help us to check if the collapsing is not happening or if we need to add a new rule to control a specific URL.
Thanks for the information, Hector. I do see the values returned for the Server Requests.
How do we know what value relates to 'too many' server requests? Are there specific threshold values we shouuld be watching for?