Elastic agent complete (7.17.1) stops working

I have the elastic agent complete (7.17.1) docker running on AWS ECS (Faregate). After a few days of running, all monitoring are failing and no steps are executed.
Did anyone have similar experience with that? Is there a way to set how many concurrent monitors to run? (I know it can be done with he heartbeat but how to set it with elastic agent complete docker image)

Hello @Tzachi.shachar Welcome to elastic Discuss forum.

Do you see any logs in the agent, on why it's crashing? I will try to figure out the concurrent monitors limit option in elastic-agent complete and will let you know if it's possible.

I guess it's crashing because of not enough resources? Does increasing resources helps?



So i checked with team, and it seems like there is no options to set concurrency right now when running things via fleet/Elastic agent complete. But we are trying to improve things , which can be tracked in these issues [Heartbeat] Check for free memory before starting synthetic browser · Issue #23687 · elastic/beats · GitHub and [Heartbeat] Automatically set browser concurrency based on memory · Issue #32082 · elastic/beats · GitHub.

That being said, i think default memory in AWS is very low, increasing it should improve things for you.


Hi @shahzad31

Thanks for your answer.
AWS was configured with 4gb memory which is not the default and should be enough to run 1 integration at a time ( there are 4 in total and they don’t run at the same time).
He fact that it was running for a few days with no issues only that the memory usage is increasing, probably means there is a leak so increasing the memory allocated at AWS will probably buy me some time but it will stop again. There are no logs as it is running on docker on aws just the metrics that indicate the cpu is near 100% and the memory as well.
All integrations are shown in uptime as failed (red) but no steps were executed.
Happy to try any ideas you might have.


@Tzachi.shachar sorry to hear about these issues. Can you share any charts or graphs of memory use? Seeing memory use over time would be quite helpful. Is it increasingly slowly and regularly, or is it spikey? If you have a process level view into which processes in the container are using what memory that would also be helpful.

Finally, given that browsers are a beta product, if you could try again on the latest 8.3.1 version that would be helpful has well.

This topic was automatically closed 24 days after the last reply. New replies are no longer allowed.