PHP-FPM could crash when timeout is configured

[PierreF]

Relevant telegraf.conf

[[inputs.phpfpm]]
  urls = ["fcgi://127.0.0.1:9008/status"]
  timeout = "5s"

Logs from Telegraf

$ ./telegraf -test -config telegraf.conf 
2024-04-17T09:23:12Z I! Loading config: telegraf.conf
2024-04-17T09:23:12Z I! Starting Telegraf unknown brought to you by InfluxData the makers of InfluxDB
2024-04-17T09:23:12Z I! Available plugins: 234 inputs, 9 aggregators, 32 processors, 25 parsers, 60 outputs, 5 secret-stores
2024-04-17T09:23:12Z I! Loaded inputs: phpfpm
2024-04-17T09:23:12Z I! Loaded aggregators: 
2024-04-17T09:23:12Z I! Loaded processors: 
2024-04-17T09:23:12Z I! Loaded secretstores: 
2024-04-17T09:23:12Z W! Outputs are not used in testing mode!
2024-04-17T09:23:12Z I! Tags enabled: host=MBP-de-pierre
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x2 addr=0x38 pc=0x1053632f4]

goroutine 116 [running]:
github.com/influxdata/telegraf/plugins/inputs/phpfpm.newFcgiClient(0x12a05f200, {0x14001bc23c8?, 0x9?}, {0x14000de6708?, 0x0?, 0x0?})
        /Users/pierref/dev/telegraf/plugins/inputs/phpfpm/fcgi_client.go:36 +0x1f4
github.com/influxdata/telegraf/plugins/inputs/phpfpm.(*phpfpm).gatherServer(0x14000edaa00, {0x14001bc23c1, 0x1c}, {0x10b7c8d40, 0x14000ee62a0})
        /Users/pierref/dev/telegraf/plugins/inputs/phpfpm/phpfpm.go:162 +0x24c
github.com/influxdata/telegraf/plugins/inputs/phpfpm.(*phpfpm).Gather.func1({0x14001bc23c1?, 0x0?})
        /Users/pierref/dev/telegraf/plugins/inputs/phpfpm/phpfpm.go:129 +0x70
created by github.com/influxdata/telegraf/plugins/inputs/phpfpm.(*phpfpm).Gather in goroutine 115
        /Users/pierref/dev/telegraf/plugins/inputs/phpfpm/phpfpm.go:127 +0x68

System info

Telegraf master, MacOS 14.4

Docker

No response

Steps to reproduce

1. Do NOT run php-fpm, have nothing listening on port 9008
2. Run Telegraf --test

Expected behavior

Don't crash :)

Actual behavior

Crash

Additional info

🤦 I've caused this issue in #15036 because I've never tested with no php-fpm running.

I'll submit a PR to fix in the next hours.

[srebhan]

@PierreF please test the binary in PR #15179, available as soon as CI finished the tests, and let me know if this fixes the issue!

[PierreF]

I've tested and it works with both #15179 and #15176 (before my first push force):

$ ./telegraf -test -config ../telegraf.conf
2024-04-17T13:30:50Z I! Loading config: ../telegraf.conf
2024-04-17T13:30:50Z I! Starting Telegraf 1.31.0-35085cc3 brought to you by InfluxData the makers of InfluxDB
2024-04-17T13:30:50Z I! Available plugins: 234 inputs, 9 aggregators, 32 processors, 25 parsers, 60 outputs, 5 secret-stores
2024-04-17T13:30:50Z I! Loaded inputs: phpfpm
2024-04-17T13:30:50Z I! Loaded aggregators:
2024-04-17T13:30:50Z I! Loaded processors:
2024-04-17T13:30:50Z I! Loaded secretstores:
2024-04-17T13:30:50Z W! Outputs are not used in testing mode!
2024-04-17T13:30:50Z I! Tags enabled: host=MBP-de-pierre
2024-04-17T13:30:50Z E! [inputs.phpfpm] Error in plugin: dial tcp 127.0.0.1:9008: connect: connection refused
2024-04-17T13:30:50Z E! [telegraf] Error running agent: input plugins recorded 1 errors

(Or 2024-04-17T13:30:27Z I! Starting Telegraf 1.31.0-1a584ffe brought to you by InfluxData the makers of InfluxDB)