https://pulumi.com logo
Title
s

shy-oxygen-8874

03/10/2021, 1:28 AM
Hi, I just ran a pulumi up that reported that it succeeded, but one resource in the stack was not created. I never got an error at any point, and it wasn't until a test ran later that I caught it at all. A pulumi preview --refresh showed that the resource needed to be created. This is the second time I've encountered this with Pulumi. I don't remember what the resource was the first time (I think it was a VPC SG, but I'm not sure). This time it was an alb listener rule. Is there some setting to make pulumi more sensitive to these failures?
l

little-cartoon-10569

03/10/2021, 1:30 AM
Did the up say it had been created, or did it entirely "miss" that it needed to be created?
s

shy-oxygen-8874

03/10/2021, 1:31 AM
It said it had been created
+  aws:elasticloadbalancingv2:ListenerRule prod-web-listener-rule created
l

little-cartoon-10569

03/10/2021, 1:33 AM
ListenerRules aren't heavyweight.. it should create it immediately... odd.
As to your question: no, other than increasing the log level (e.g.
-v=9
) I don't think there's a way to be "more aware" of provisioning failures.
s

shy-oxygen-8874

03/10/2021, 1:35 AM
It's troubling--if the resource isn't there I'd expect to get a failure.
l

little-cartoon-10569

03/10/2021, 1:35 AM
Yep. There are some sorts of resources where I accept that the request to create can be successful, and the actual creation might take a while and later be unsuccessful.. that's just clouds for you..
s

shy-oxygen-8874

03/10/2021, 1:36 AM
Yeah, I suppose, to a point. Although, the way CF handles that is to have a long poller for status
If you're counting on your deployment status for CI/CD, you'd hope there'd be a stronger guarentee
s

sparse-park-68967

03/10/2021, 1:59 AM
Could you cut a ticket with the details? As close a representation of your code as is possible would be great.