Hi All, I’ve noticed sometimes a run will fail bec...
# general
Hi All, I’ve noticed sometimes a run will fail because a back-end cloud validation error occurs but this is not always passed back to Pulumi in realtime… the error only appears after a timeout period is reached or the program is terminated early. Is there any plan to observe the state of operations in realtime?
Hey Danno, I’m not 100% sure I understand. Could you give an example? I’m not aware of any plans to allow observing provider behavior beyond the output of
pulumi up
with logs. We’re happy to take suggestions.
I was provisioning a nat gateway and my eip limit was reached - it took a long time to error out…
Thanks for clarifying. It sounds like that kind of observability would require reaching into a provider to see what it is doing. You could experiment with running
with higher log levels enabled. That will get you more insight into the provider’s operations.
thx. I remember there was an experimental pulumi observation feature - which would attempt to tail events from cloudtrail… i’m not sure it ever worked fully tho
If i had to take an architectural guess… i’d say that Pulumi monitors the resource which it’s currently provisioning for status… but if this resource has an upstream resource dependency (like natgw depends on an eip) then the error returned from eip is not necessarily propagated back to the natgw api directly… which means we have to wait for the provisioning timeout… @billowy-army-68599 does this sound plausible?