prometheus-to-cloudwatch icon indicating copy to clipboard operation
prometheus-to-cloudwatch copied to clipboard

Http connections not being closed

Open Codexxl opened this issue 5 years ago • 4 comments

We're running prometheus-to-cloudwatch in a kubernetes pod to scrape our selenium endpoint for metrics. This is working but after a couple of hours it causes the selenium exporter to become unavailable as the http connections are being held open so no new ones can be created. We're having to restart the exporter to get everything working again. Is there a way to get prometheus-to-cloudwatch to close the http connections after it has made them so that it isn't leaving them hanging?

Codexxl avatar Oct 31 '19 12:10 Codexxl

this is what I'm struggling as well. Any workaround?

jjuraszek avatar Jun 04 '20 08:06 jjuraszek

@jjuraszek We had to resort to a cron job that restarted our exporter every couple of hours so that we could force close the connections.

Codexxl avatar Jun 04 '20 08:06 Codexxl

This open PR https://github.com/cloudposse/prometheus-to-cloudwatch/pull/39 seems to fix the problem.

dcwangmit01 avatar Jun 10 '20 21:06 dcwangmit01

Now that #39 is merged, can this issue be closed?

carljohnstone avatar Sep 23 '20 14:09 carljohnstone