Telegraf crashes with the following error: panic: runtime error: invalid memory address or nil pointer dereference

[placyd]

I was running telegraf 1.18.0-alpine on docker container on Azure App Service.

Relevant telegraf.conf:

[agent]
  interval = "20s"
  round_interval = true
  metric_batch_size = 1000
  metric_buffer_limit = 10000
  collection_jitter = "0s"
  flush_interval = "20s"
  flush_jitter = "0s"
  precision = ""
  hostname = ""
  omit_hostname = false
  debug = true


[[outputs.azure_monitor]]
  region = "northeurope"
  resource_id = "/subscriptions/*******/resourceGroups/*****/providers/Microsoft.Web/sites/****"
  namespace_prefix = "Telegraf_app/"

[[inputs.prometheus]]
  urls = ["https://*****.azurewebsites.net/metrics"]
  metric_version = 2

System info:

Telegraf in docker container: version: 1.18.0-alpine

Steps to reproduce:

launch telegraf

Expected behavior:

Sending logs to Azure monitor

Actual behavior:

Telegraf crashed.

Additional info:

logs:

- Container for ***** site ***** is unhealthy, Stopping site.
- Stopping site ***** because it is not healthy.
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x18 pc=0x22ebdd5]
goroutine 114 [running]:
github.com/influxdata/telegraf/plugins/outputs/azure_monitor.(*AzureMonitor).send(0xc00022b760, 0xc0004f0000, 0x21c7, 0x2800, 0x0, 0x0)
/go/src/github.com/influxdata/telegraf/plugins/outputs/azure_monitor/azure_monitor.go:348 +0x335
github.com/influxdata/telegraf/plugins/outputs/azure_monitor.(*AzureMonitor).Write(0xc00022b760, 0xc0000bd900, 0x26, 0x26, 0xc0000bd900, 0xc000535ee0)
/go/src/github.com/influxdata/telegraf/plugins/outputs/azure_monitor/azure_monitor.go:325 +0x5fa
github.com/influxdata/telegraf/models.(*RunningOutput).write(0xc0000ea900, 0xc0000bd900, 0x26, 0x26, 0x26, 0xc0003e7801)
/go/src/github.com/influxdata/telegraf/models/running_output.go:246 +0x99
github.com/influxdata/telegraf/models.(*RunningOutput).Write(0xc0000ea900, 0x86d5000, 0xf6000)
/go/src/github.com/influxdata/telegraf/models/running_output.go:203 +0x1f8
github.com/influxdata/telegraf/agent.(*Agent).flushOnce.func1(0xc00049c3c0, 0xc000bb9290)
/go/src/github.com/influxdata/telegraf/agent/agent.go:843 +0x27
created by github.com/influxdata/telegraf/agent.(*Agent).flushOnce
/go/src/github.com/influxdata/telegraf/agent/agent.go:842 +0x7a

[ivorybilled]

Hi, is this happening when telegraf is shutting down and/or right after an occurs with the output plugin? Otherwise, I suspect based on the stack trace and a small oversight I just found (soon to be fixed), that your 'auth' object isn't being created properly, and ends up being nil.

We aren't handling the error if an error occurs while creating the auth object so it won't currently show what the issue is, but for now I'd suggest checking up on your authorization / environment variables that are used for azure auth to see if there might be any obvious issues. Let me know if this doesn't sound right.

[ivorybilled]

Hi @placyd, if you'd like, could you grab a build from the pull request above and test if your panic still occurs?

[placyd]

@jagularr Thank you for the quick improvement. I think we'll test it on Monday.

[placyd]

@jagularr looking on the fix I have no idea why but now Azure Monitor output plugin started working as expected. Thank you for your effort. I wish this fix will goes to official release soon.