Dashboard quick ranges only show as many processes as minutes in time(), despite more data points from more processes

Hey there,

Thanks for looking at this.

Here’s the dashboard query:

SELECT mean("cpu_usage") FROM "procstat" WHERE "host" =~ /^(HOSTNAMESNIPPED)$/ AND time > 1524380400000ms and time < 1524985199999ms AND "process_name" =~ /^php-fpm.*/ GROUP BY time(5m), "process_name" fill(null)

Every quick range from “last 5 minutes” to “today so far” produces expected graph results. Any other range lists only as many processes as minutes in time(), so if I set time(6m) it displays the graphs of 6 processes at 6m intervals. To display all of the processes I would need to raise the interval to 30m or in some cases more.

Following the “how to troubleshoot metric query issues” i noticed that

partial":true}],"partial":true

Is being attached to results. Does this mean the data is being cut short because there’s too much?

Any advice is appreciated. Will be actively monitoring.

Googling the partial true led me this github issue:

This led me to set max-row-limit to zero in influxdb:

influxdata/influxdb#7879

Now my graphs are slow, but they work, which is a very welcome trade off.