x509_cert input: certificate validation is not working as expected

[the-smooth-operator]

Relevant telegraf.conf:

[[outputs.file]]

[[inputs.x509_cert]]
  sources = [
    "https://sha512.badssl.com:443",
    "https://expired.badssl.com:443",
    "https://preact-cli.badssl.com:443",
    "https://google.com:443",
  ]

System info:

Built from master rev. b9cb606
Linux.

Steps to reproduce:

1. paste that configuration into a file
2. build Telegraf from master and run ./telegraf --config config.conf --test

Expected behavior:

I don't see any verification_error

Actual behavior:

The parsing is reporting first an invalid cert, and later a valid one. This happens with all the sites I tested.

> x509_cert,common_name=*.badssl.com,country=US,host=banbao,locality=Walnut\ Creek,organization=Lucas\ Garron,province=California,source=https://sha512.badssl.com:443,verification=invalid age=74079777i,enddate=1585137600i,expiry=21262622i,startdate=1489795200i,verification_code=1i,verification_error="x509: certificate signed by unknown authority" 1563874978000000000
> x509_cert,common_name=DigiCert\ SHA2\ Secure\ Server\ CA,country=US,host=banbao,organization=DigiCert\ Inc,source=https://sha512.badssl.com:443,verification=valid age=201130977i,enddate=1678276800i,expiry=114401822i,startdate=1362744000i,verification_code=0i 1563874978000000000

> x509_cert,common_name=www.google.com,country=US,host=banbao,locality=Mountain\ View,organization=Google\ LLC,province=California,source=https://www.google.com:443,verification=invalid age=3028799i,enddate=1568103300i,expiry=4228322i,startdate=1560846178i,verification_code=1i,verification_error="x509: certificate signed by unknown authority" 1563874978000000000
> x509_cert,common_name=Google\ Internet\ Authority\ G3,country=US,host=banbao,organization=Google\ Trust\ Services,source=https://www.google.com:443,verification=valid age=66390135i,enddate=1639526442i,expiry=75651464i,startdate=1497484842i,verification_code=0i 1563874978000000000

Additional info:

The feature was merged yesterday on #6143

[faabsen]

It seems like the bug is stillt around, after using the image (v1.18.0/v1.18.1) from docker hub I can reproduce similar errors:
depending on how the URLs are specified (the order is important) there are different, faulty results. Anyone having a similar behavior?