Unhelpful error message when attempting to re-run failed create about bosh-bootloader HOT 7 CLOSED

We have created an issue in Pivotal Tracker to manage this. You can view the current status of your issue at: https://www.pivotaltracker.com/story/show/117271935.

from bosh-bootloader.

I seem to get this failure even when the step that fails is after the cloudformation step (e.g. the bosh-init deploy fails). Every time I have to delete the cloudformation stack manually in order to run the tool again.

from bosh-bootloader.

@robdimsdale one current limitation of bbl, is that we only write out the state.json when the command successfully finishes, so when your command failed, it left AWS artifacts which are not tracked by bbl. Also, the destroy command needs stack and bosh information in order to do it's job, so your destroy failed because that information didn't exist.

We have a track of stories for bbl to gracefully handle errors during a provision, but until we get to that point, it may be worth providing a more useful error message letting the user know that they should delete the stack which bbl created.

from bosh-bootloader.

Hey @robdimsdale

Do you have any events from the AWS console explaining why the original cloudformation stack might have failed?

We have stories that would fix the destroy problem you hit. But we should still improve the original error message.

Instead of:

aws cloudformation failed: ROLLBACK_COMPLETE

It should say:

Failed creating cloudformation stack `STACK_NAME'.
Check the AWS console for error events related to this
stack, and/or open a GitHub issue at
https://github.com/pivotal-cf-experimental/bosh-bootloader/issues.

from bosh-bootloader.

No logs anymore, but here's what I think happened:

• run bbl first time - bosh-init step fails due to nokogiri / xcode
• install xcode and nokogiri, re-run bbl - I'm not quite sure what happened here, but I think bbl had no state file so tried to create an entirely new stack.
  • This failed because I reached my VPC limit. I remember seeing two stacks in Cloudformation.
• I deleted both stacks (including having to destroy the bosh-init director manually from the first stack as it wasn't part of Cloud Formation) and ran bbl again and it succeeded.

from bosh-bootloader.

Yes, re-running bbl created a new stack because it's not re-entrant yet. This epic will address that.

from bosh-bootloader.

Ok, well it sounds like we understood what went wrong and have stories/epics to address the various pain points along the way. I'll close this issue now. Thanks for the updates and explanations.

from bosh-bootloader.