pg_up metric 0 value is absent in query return when pg service is down

1. start stage instance with PG and your ssh key
2. access stage via ssh
3. execute "pmm-admin list" and get PG_SERVICE_NAME
4. execute "docker stop {PG_SERVICE_NAME}"
5. in pmm admin navigate to explore
6. run query "pg_up{service_name='PG_SERVICE_NAME'}"

Expected result: pg_up 0 and the graph shows line on 0 level;
Actual result:  pg_up is null, and graph line ends on level 1
 
 
Additionally:

in "pmm inventory > agents" tab there is an agent id and  port  for a target Postresql Exporter. Execute to see that metric is there: 

curl --silent -u "pmm:/agent_id/xxx" 'http://stage_ip:port/metrics' | grep 'pg_up'

Workaround from anton.bystrov is to increase custom metric resolution "high" interval to 10 seconds.

see automated test: https://github.com/percona/pmm-ui-tests/blob/PMM-9384-part2/tests/metrics/verifyPostgresql_test.js#L29