19 Nov 2020 11:37 AM
Hi,
What is the impact on the enduser experience with (Mobile and) Agentless RUM with javascript agent, when the Cluster Active Gate (on prem in Managed environment) becomes unavailable? Will there be a noticeable delay?
And if in DNS two CAGs are defined for the endpoint URL, and one of them is available, will there be a, noticeable, timeout trying the first/cached IP, before trying and connection to the next? Seeing if DNS Round Robin mechanism is suitable for a prospect.
Solved! Go to Solution.
20 Nov 2020 08:27 AM
Agentless RUM sends results as beacons as async non-blocking operation to a defined endpoint. Endpoint is resolved by a DNS to a single IP address. For that reason a load balancer is recommended to ensure data is always transferred to a working Cluster AG.
If the endpoint is unavailable an event/data is lost. However, by no means it impacts on how website performs (just no monitoring data). Additionally for Mobile RUM there are retry attempts for 10 minutes. After that time monitoring data is lost.
Explaining that, DNS Round Robin will partially solve the issue - however you might observe gaps in data when an unreachable IP is returned by DNS.
20 Nov 2020 10:41 AM
Thank you Radowslaw, especially that async non-blocking was a concern.
05 Dec 2020 12:10 PM
Hi @Radoslaw S., I like to follow up on this, because I think there is a situation where an unreachable AG causes delay in the browser;
As I understand there are 4 RUM JavaScript injection methods. If the Code snippet synchronous (syncCS) method is used, this has impact, it is noted as render blocking, and tested in lab as such, with firewall blocked AG. For external purposes this method likely should be used with more care, and asynchronized methods are preferred, am I right?
RUM JavaScript injection | Dynatrace Help
07 Dec 2020 09:49 AM
That is true. You previously asked about sending results data. This one is about actually getting the configuration - for which you have different methods to select.