External-dns >0.15.1 failing with nil pointer dereference on startup

[JyothsnaGuduri]

What happened:
After upgrading external-dns to version 0.15.1(helm version 1.15.1) from 0.15.0, app showed following errors and failed to come up healthy

panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x8 pc=0x25e0d89]

goroutine 1 [running]:
sigs.k8s.io/external-dns/provider/awssd.(*AWSSDProvider).instancesToEndpoint(0xc000d692c0?, 0x4599c58?, 0xc000421110, {0xc000d96ac0, 0x1, 0x10?})
sigs.k8s.io/external-dns/provider/awssd/aws_sd.go:170 +0x89
sigs.k8s.io/external-dns/provider/awssd.(*AWSSDProvider).Records(0xc0006e61a0, {0x4599c58, 0xc0004e3040})
sigs.k8s.io/external-dns/provider/awssd/aws_sd.go:158 +0x350
sigs.k8s.io/external-dns/registry.(*AWSSDRegistry).Records(0xc00022cb40?, {0x4599c58?, 0xc0004e3040?})
sigs.k8s.io/external-dns/registry/aws_sd_registry.go:56 +0x29
sigs.k8s.io/external-dns/controller.(*Controller).RunOnce(0xc000a58b40, {0x4599c58, 0xc0004e3040})
sigs.k8s.io/external-dns/controller/controller.go:212 +0xf8
sigs.k8s.io/external-dns/controller.(*Controller).Run(0xc000a58b40, {0x4599c58, 0xc0004e3040})
sigs.k8s.io/external-dns/controller/controller.go:360 +0xc5
main.main()
sigs.k8s.io/external-dns/main.go:434 +0x396f

What you expected to happen:
App should come up healthy without any errors.

How to reproduce it (as minimally and precisely as possible):
We are using following config

registry: aws-sd
provider:
  name: aws-sd
policy:sync

We have AWS cloudMap domain setup with one of the record not having description defined.

Anything else we need to know?:

Environment:

• External-DNS version (use external-dns --version): >0.15.1
• DNS provider: AWS cloudmap
• Others:

[ivankatliarchuk]

Hi.

Please note that we do not offer patch support for released versions. However, we have implemented numerous fixes in the registry following release 0.15.1 (see linked pull requests)

We also strongly recommend trying the latest version, 0.16.1. If the problem still exists, make sure to share all the arguments (not helm).

Apologies we do not support, or there is no way to push patches to released versions. Since release 0.15.1 we released multiple pathces to the registry

Any chance you share all arguments. Is it possible to try latest version which is 0.16.1?

The most recent pull-request, we did some smoke tests

Pathches to aws-sd registry https://github.com/kubernetes-sigs/external-dns/pulls?q=is%3Apr+aws-sd+is%3Aclosed

0.16.1 https://github.com/kubernetes-sigs/external-dns/releases/tag/v0.16.0

Smoke tests from latest patch related to aws-sd

• fix(aws-sd): service instances registration and deregistration #5135 (comment)
• fix(aws-sd): service instances registration and deregistration #5135 (comment)

[restlesswind]

we verified that the problem is caused by the following line not handling CloudMap records with empty description and the code hasn't been fixed in the later or latest releases.

labels[endpoint.AWSSDDescriptionLabel] = *srv.Description

https://github.com/kubernetes-sigs/external-dns/blob/v0.15.1/provider/awssd/aws_sd.go#L170

[ivankatliarchuk]

I might explain differently. If the bug is in version 0.15.1 but not in 0.16.1, nothing we could do. There seems to be an bug aka null pointer which is not handled in 0.15.1 version.

What we need
1. confirmation that the issue is in 0.16.1
2. terraform, aws cli or anything else to create this CloudMap records. I'm not using cloudmap, so no idea how to create records.

If this cannot be provided, I would prefer to pass it on to some other volunteer who have time to guess and try reproducing it.

[JyothsnaGuduri]

1. Yes, I can confirm the issue exists in 0.16.1.
2. We are using terraform to create one of the cloudmap record that doesn't have description for the record. All others are managed by ingress and external-dns service.

We are using most of the default values except for the following

interval: 10m
triggerLoopOnEvent: true
policy: sync
txtOwnerId: ""<can't share the exact value>
domainFilters: [] <can't share the exact value>
registry: aws-sd
provider:
  name: aws-sd
extraArgs: 
  - --aws-zone-type=private
  - --aws-zones-cache-duration=1h
  - --aws-batch-change-size=6000
  - --aws-batch-change-interval=10s