Nil inhibit rule in alertmanager configuration causing alertmanager pods to crash because of NPE

[rohang98]

I was running Cortex and I came across an issue that caused my alertmanager pods to crash unexpectedly. I was trying to use the following alertmanager configuration:

global:
  resolve_timeout: 5s
receivers:
  - name: example-email
    email_configs:
      - to: 'youraddress@example.org'
inhibit_rules:
  -

I forgot to add my inhibition rules but didn't realize it would cause a NPE. The error from the crash was:

panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x10a83b1]

goroutine 261 [running]:
golang.a2z.com/AWSPrometheusCortex/vendor/github.com/prometheus/alertmanager/inhibit.NewInhibitRule(0x0)

The error is coming from https://github.com/prometheus/alertmanager/blob/main/inhibit/inhibit.go#L53-L56:

	for _, cr := range rs {
		r := NewInhibitRule(cr)
		ih.rules = append(ih.rules, r)
	}

Are nil list elements, in this case inhibition rules, processed even if they are nil on purpose? If I add if cr == nil continue to the code, it doesn't cause the alertmanager pods to crash since it skips the nil inhibition rule.

[simonpasquier]

Thanks for the report! Not sure why it wasn't noticed before but it should be fixed.

Another null-pointer panic would also happen with this snippet:

receivers:
-
- name: foo

I think that we should change the slices of pointer items to slices of value items.

alertmanager/config/config.go

Lines 300 to 301 in 9aa5974

	InhibitRules []*InhibitRule `yaml:"inhibit_rules,omitempty" json:"inhibit_rules,omitempty"`
	Receivers []*Receiver `yaml:"receivers,omitempty" json:"receivers,omitempty"`

cc @gotjosh

[rohang98]

Thanks for the comment! I'll work on the PR to address this.