Please check failing tests on FB and return to QA once all is green, thanks.

It looks like 1 second offset hasn't always enough.
I have increased it to 3 seconds and tested on the provided installation.

 I can still see this problem when changing time range - the last interval shows as DOWN even if agent is running. Please check http://3.21.240.37/graph/d/vmagent/victoriametrics-agents-overview?viewPanel=13&orgId=1&refresh=1m&from=now-6h&to=now&var-interval=$__auto_interval_interval&var-ds=Metrics&var-node_name=ip-10-178-1-233.us-east-2.compute.internal&var-instance=All&var-job=All 

Thanks, ! this helps! was able to see some problems
https://pmmdemo.percona.com/graph/d/vmagent/victoriametrics-agents-overview?viewPanel=13&folder=current&orgId=1&from=1629178631333&to=1629185153140&var-interval=1s&var-ds=Metrics&var-node_name=ps-80-source&var-instance=All&var-job=All

Looks like the problem is related to "incomplete interval" or some cases when the last 1 sec data is not processed correctly. 

, PTAL.. I think there was some better solution for this problem from VM...

Hi  -

I was able to view this condition on PMM Demo.  I went to the same VictoriaMetrics Agents dashboard and kept changing the time ranges on the Uptime visualization and was able to see it.