Unexpected schema changes while using aggregateWindows fn: max

Grafana Time Series Panel
1

Hi everyone, I’m new to Grafana and InfluxDB, but recently I came across a problem with aggregateWindow function and solved the issue by sheer luck. It’d be greatly appreciated if anyone could provide an explanation why it works.
Thank you in advance!

  • What Grafana version and what operating system are you using?
    Grafana v10.3.3 on Ubuntu 22.04
    InfluxDB v2.7.5

  • What are you trying to achieve?

  1. To create a Time series graph for CPU utilization of multiple VMs using aggregateWindow function to show max value (fn: max) while dynamically group data in a “readable” manner.
    Note: data is polled every 20 seconds, from vCenter to InfluxDB using Telegraf.

Flux query:
from(bucket: v.defaultBucket)
|> range(start: v.timeRangeStart, stop: v.timeRangeStop)
|> filter(fn: (r) => r[“_measurement”] == “vsphere_vm_cpu”)
|> filter(fn: (r) => r[“_field”] == “usage_average”)
|> filter(fn: (r) => r[“cpu”] == “instance-total”)
|> filter(fn: (r) => r[“vmname”] =~ /some regex)
|> group(columns: [“vmname”])
|> aggregateWindow(every: v.windowPeriod, fn: max, createEmpty: false)
|> yield(name: “max”)

  • What happened? / What’s wrong?
    Result:
  1. fn: max causes field/ column names mixed with start, stop and vmname label tags. E.g. _value {_start=###, _stop=###, vmname=###}.
  2. All labels/ tags are truncated into individual fields/ columns. If fn: mean, there are only 2 fields: time and value, each series is separated by VM name.
  • What did you expect to happen?
  1. Only time field named “time” and value field named by the VM names.
  2. Max value of the interval should be displayed, instead of taking the mean of the interval which could result in lower peak value/ lower max value in the graph.
  3. Even with a larger timeframe, say last 7 days, the data points should be visible and readable. So aggregateWindow is really important in my case, without this function, there will be too many data points.
  • What else did I try? (details to be shared upon request)
  1. Map functions to rename field + select only wanted fields. I tried map using r with r[“vmname”] but no luck. I did not find another way to put r[“vmname”] as series key(?), and column name.
  2. Data transformation to replace field that matches (.){.} by $1. It reduced the field/ column name to “_value” but data can no longer be distinguished by VM.
  3. In grafana options, set “Display name” to variable $vm but it will retrieve entire list of VM instead of specific VM.
  4. No group columns and transform field names into variable, unfortunately, transformation does not allow reference (using variable $vm).

Flux query:
from(bucket: v.defaultBucket)
|> range(start: v.timeRangeStart, stop: v.timeRangeStop)
|> filter(fn: (r) => r[“_measurement”] == “vsphere_vm_cpu”)
|> filter(fn: (r) => r[“_field”] == “usage_average”)
|> filter(fn: (r) => r[“cpu”] == “instance-total”)
|> filter(fn: (r) => r[“vmname”] =~ /some regex/)
|> group(columns: [“vmname”])
|> aggregateWindow(every: v.windowPeriod, fn: max, createEmpty: false)
|> aggregateWindow(every: v.windowPeriod, fn: mean, createEmpty: false)
|> yield(name: “mean”)

I found this solution after a week of trial and error, could anyone shed some light? is the fn: max behavior worked as expected?
I tried both queries on InfluxDB Data Explorer (Query/ Script builder) but got no problems with X/ Y columns. It only happens on Grafana.
I’d really appreciate any clues/ answers. Many thanks!

2

Solution:

from(bucket: v.defaultBucket)
|> range(start: v.timeRangeStart, stop: v.timeRangeStop)
|> filter(fn: (r) => r[“_measurement”] == “vsphere_vm_cpu”)
|> filter(fn: (r) => r[“_field”] == “usage_average”)
|> filter(fn: (r) => r[“cpu”] == “instance-total”)
|> filter(fn: (r) => r[“vmname”] =~ /some regex/)
|> group(columns: [“vmname”])
|> aggregateWindow(every: v.windowPeriod, fn: max, createEmpty: false)
|> aggregateWindow(every: v.windowPeriod, fn: mean, createEmpty: false)
|> yield(name: “mean”)

1 Like