I created a Kubernetes deployment on Azure (AKS, H...
# general
g
I created a Kubernetes deployment on Azure (AKS, Helm Charts and various other resources - all created by Pulumi). “pulumi up” timeouted on a couple of things (ingress, a deployment with a crash-loop, and another deployment). If I run “pulumi refresh” it shows “0 changes”. If I run “pulumi up” it wants to create those 3 components. The problem is, that the components do exist actually, and when Pulumi wants to create them, it fails with “error: Plan apply failed: ingresses.extensions “xyz” already exists”.
c
@gray-city-50684 can you look in the package lock and tell me what version pulumi/kubernetes resolved to?
g
0.17.3
c
@gray-city-50684 sorry, I mean can you show me the completely resolved version. Like this:
Copy code
"@pulumi/pulumi@^0.15.1":
  version "0.15.1"
  resolved "<https://registry.yarnpkg.com/@pulumi/pulumi/-/pulumi-0.15.1.tgz#1b94077ec0bfb8fef868d6aa66643d7f726c0678>"
g
@echoing-match-29901/kubernetes”: { “version”: “0.17.3”, “resolved”: “https://registry.npmjs.org/@pulumi/kubernetes/-/kubernetes-0.17.3.tgz“, “integrity”: “sha512-bCLLNWWfs81FY/y5wnAP2gVE3WcCJfzRB8U8qbyoM3IfRVE7vwjn0hHzPxE+HkEHgEVERQMAWBNaWWCo62gLdA==“, “requires”: { “@echoing-match-29901/pulumi”: “^0.15.1”, “@types/glob”: “^5.0.35”, “@types/js-yaml”: “^3.11.2”, “@types/mocha”: “^5.2.5”, “@types/tmp”: “^0.0.33”, “glob”: “^7.1.2”, “js-yaml”: “^3.12.0”, “mocha”: “^5.2.0”, “shell-quote”: “^1.6.1”, “tmp”: “^0.0.33” }, “dependencies”: { “@echoing-match-29901/pulumi”: { “version”: “0.15.4”, “resolved”: “https://registry.npmjs.org/@pulumi/pulumi/-/pulumi-0.15.4.tgz”, “integrity”: “sha512-7ZBEZH5SuaYuIm06pH8pS8VPWeISGOkC4wjsEbbNC44RD7a4LCWcvIsx8Avbdi1fPR4hNqmL6Tp/nceJf4RzZw==“, “requires”: { “google-protobuf”: “^3.5.0”, “grpc”: “^1.12.2”, “minimist”: “^1.2.0”, “normalize-package-data”: “^2.4.0”, “protobufjs”: “^6.8.6”, “read-package-tree”: “^5.2.1”, “require-from-string”: “^2.0.1”, “source-map-support”: “^0.4.16”, “ts-node”: “^7.0.0”, “typescript”: “^3.0.0” } } } },
c
@gray-city-50684 thanks. That’s very likely a bug.
Do you have a repro?
This should be fixed.
g
I didn’t want to destroy yet, wanted to preserve the state
I can try it again
btw: I had another strange thing happen: on the first run of “pulumi up” when I created the AKS cluster, it failed with error on can’t find ServicePrincipal
c
that, unfortunately, is pretty common, and also an azure issue
g
the ServicePrincipal is also created by the script, and I never had a problem with like this
ah, ok
c
yeah
g
any know workarounds?
c
awhile ago I got 500s from azure when I tried to create service principal.
You could create the resource after a
setTimeout
🙂
g
🙂 ok
c
or something like that.
I’m trying to repro right now.
g
maybe the easiest would be to have a Deployment that crash-loops, as one of mine did
(some new setting in AKS RBAC, that the default service account can’t monitor ConfigMaps)
c
sorry got distracted by a teammate
I’m going to try exactly that.
@gray-city-50684 I can’t reproduce if I press ctrl-c
I’ll try now to wait for the timeout.
should take 10 minutes
g
ok, I’ll have to leave work in about 50 minutes - but I’ll do work tomorrow, so I’ll try to reproduce, if it doesn’t happen for you now
c
@gray-city-50684 reproduced. Will have a fix out in an hour or two.
g
cool, thx
I’ll try and test it tomorrow
c
@gray-city-50684 fixed.
we’ll roll the release later todya.