bummer, was hoping to start using this. project is...
# general
b
bummer, was hoping to start using this. project is awesome though Thanks for the info
c
@busy-umbrella-36067 I’m curious to hear what the problem is
is the issue just that you don’t want the state to be in the Pulumi service?
b
I dont want deployment to not work if the cloud service is down
everyone here is already using terraform with s3 remote state
dont wanna talk security into allowing yet another oauth github app
c
What do you mean by having S3 support “built in” then?
b
stacks stored on s3
with locking
c
That would still include a service, no?
Locking is provided by the service.
That issue tracks whether the statefile is stored in Pulumi or in the service’s database, right?
b
yes
c
In general to provide locking some service would have to be involved, right?
b
basically, if you go under. we still need to be able to work in a team env
c
Right, that should work with the local backend.
You can just make your CI/CD solutino that invokes
pulumi up
be the lock, right?
b
with the local backend, I have to git recomit and push
c
hmm, not sure I understand.
b
to share the stack with other people
state*
c
Why not just have your CI/CD scripts push/pull from s3?
Your CI/CD is going to “lock” anyway, right?
b
because we also need the ability to apply locally
c
Ah.
I see.
b
if it was just ci system launching it I wouldnt care so much for locking
c
Right, right. This makes sense.
This is one place where terraform apply really works out.
b
we are using that heavily for gitops right now
but the architecture of TF is a pain point right now
c
Having a separate plan/apply makes this nice.
You mean, the statefile stuff, or something else?
b
module nesting 🤮
c
ah
b
and count on modules
that just baffles me. no bool toggles for resources that are modules
c
Yeah, making a new language is hard. 🙂
💯 1
b
right, I don’t want to sound like I dont appreciate the work all the people put into it.
c
No, I feel your pain. I used to work on a system like that.