I'm seeking advice on handling large full memory crash dumps (over 100GB) from a Windows pod in Azure Kubernetes Service (AKS) after a crash. I need to ensure that the dumps are captured without corruption and can be inspected or downloaded later. Just to give you some background:
- My cluster is running on AKS.
- I've attempted using a premium Azure disk (az-disk), but it hasn't proven to be reliable for this situation.
- I'm also considering options like emptyDir, although I haven't tested that yet. Any suggestions would be greatly appreciated. Thank you!
3 Answers
You’re correct that handling memory dumps from Windows pods can be different. It might be worth considering refactoring your application to handle failures in a cleaner way. But if the crash dumps are essential, maybe ensure you’re flushing the memory correctly before trying to save it. Hope that helps!
It sounds like a tricky situation! Have you explored using a different storage option or backing up the dump to Blob Storage for safer handling? Alternatively, if you’re just testing, emptyDir might simplify things if you need something temporary. Good luck!
What’s your primary use case for the dumps? If you're looking to debug the application, knowing what kind of application you're deploying could help others tailor their advice. I’ve always found dealing with Windows containers in Kubernetes to be a real headache too!
Yeah, I feel you! Huge memory dumps can be frustrating, especially with Windows pods. Hang in there!