According to the Enterprise synthetic documentation, it is recommended that you reboot agent machines, daily. Is this really necessary or is it enough to reboot the actual agent service? The question is brought up by a customer who is concerned with issues related to the reboot of a machine.
If it is recommended, what is the reasoning behind this recommendation? And what is not achieved if you reboot the service only?
Thanks in advance!
Answer by Carol O. ·
Hi. Thanks for your question. You are correct that the doc indicates a recommended daily reboot. As stated in Agent Reboot, this is important to clean-up from the continuous running of applications. Some customers have chosen not to reboot daily if they consider their applications to not have clean-up or resource issues; others have chosen to reboot multiple times each day. Rebooting the service will not recover from any application issues themselves.
Answer by Luke B. ·
At my customer, we run about 40 agents which run the majority of our transactions, and a further 40 which are used to supplement key application where site coverage is important to them. We find rebooting the agents to be necessary, as we test a variety of web, client/server and Citrix applications. Sometimes, an application will freeze and prevent the agent from cleaning up properly. Other times, a combination of obscure application logic and our scripts will cause the agents to enter into an infinite loop (especially if you use Whenever() and the code doesn't handle the triggering event properly).
I personally find that the daily restart helps take variables out of the equation when doing our daily health status check. If the restart didn't fix the problem, there is more likely to be an actual issue that needs investigation.
Answer by Frans S. ·
Hi Ali, In our experience we prefer daily reboot, as it is recommended. Or at least once a week. Just the service IMHO is not enough. Besides daily restart can simulate daily user behavior (switching off units/logging off), the repetitive nature of transactions can cause slowness or other issues. Memory leaks/filling up, Java issues etc. Even unresponsiveness. On the other hand, a reboot does indeed have an increased cause for robot unavailability, so extra checks and measures need to be in place. One thing we also learned, is that with hardware robots, some bios settings may need to be set. Automatic Power On (to solve power off by accident for example) (at hh:mm), power on after power loss, Wake-On-Lan etc. And you may want to have a KVM-Over-IP, or PDU-over-IP (powercut) in place.
ESM Proxy Agent setup 2 Answers