https://pulumi.com logo
#general
Title
# general
i

important-leather-28796

04/03/2019, 5:48 PM
@stocky-spoon-28903 Trying to create a pulumi/gcp@0.18.2 regional cluster but encountering:
Copy code
error: gcp:container/cluster:Cluster resource 'production' has a problem: : invalid or unknown key: node_locations
    error: gcp:container/cluster:Cluster resource 'production' has a problem: : invalid or unknown key: location
s

stocky-spoon-28903

04/03/2019, 5:50 PM
Hmm, that looks like a bug.
i

important-leather-28796

04/03/2019, 5:59 PM
@stocky-spoon-28903 Is this an easy hotfix? I’m in the eleventh hour, production needs to be up tomorrow (beta starts Friday) - this was my last change.
s

stocky-spoon-28903

04/03/2019, 6:06 PM
That largely depends whether the bug is in the terraform provider or the mapping - I’ll take a look shortly
Thanks for the repro
@important-leather-28796 can you run:
Copy code
pulumi plugin ls
And post the output?
I can’t repro this - I suspect you have side by side versions of the GCP plugin and the old one is getting selected
i

important-leather-28796

04/03/2019, 7:15 PM
I’m already using ncu - so all ts dependencies are up to date
something else I need to check? brew?
0.17.4 of brew/pulumi installed
did a
yarn list @pulumi/gcp
and all are 0.18.2
double checking node_modules now
node_modules is up to date @stocky-spoon-28903 - is there somewhere else I need to check versions?
s

stocky-spoon-28903

04/03/2019, 7:18 PM
pulumi plugin ls
i

important-leather-28796

04/03/2019, 7:18 PM
wow
s

stocky-spoon-28903

04/03/2019, 7:18 PM
A good way to find out is to do:
Copy code
pulumi plugin rm --all
i

important-leather-28796

04/03/2019, 7:19 PM
Copy code
~/p/a/c/infrastructure ❯❯❯ pulumi plugin ls
NAME        KIND      VERSION  SIZE   INSTALLED    LAST USED
gcp         resource  0.18.2   64 MB  1 hour ago   now
gcp         resource  0.18.1   64 MB  1 day ago    now
gcp         resource  0.18.0   64 MB  1 week ago   now
gcp         resource  0.17.1   61 MB  3 weeks ago  now
gcp         resource  0.17.0   61 MB  3 weeks ago  now
gcp         resource  0.16.9   61 MB  4 weeks ago  now
gcp         resource  0.16.8   61 MB  1 month ago  now
kubernetes  resource  0.22.0   45 MB  1 week ago   now
kubernetes  resource  0.21.1   45 MB  1 week ago   now
kubernetes  resource  0.21.0   45 MB  3 weeks ago  now
kubernetes  resource  0.20.4   45 MB  4 weeks ago  now
kubernetes  resource  0.20.2   45 MB  1 month ago  now
random      resource  0.5.1    27 MB  3 weeks ago  now
random      resource  0.5.0    27 MB  3 weeks ago  now
random      resource  0.4.0    28 MB  1 month ago  now
random      resource  0.2.0    35 MB  1 month ago  now

TOTAL plugin cache size: 775 MB
s

stocky-spoon-28903

04/03/2019, 7:19 PM
Then remove
node_modules
, rerun
npm install
in the program, and then look at
pulumi plugin ls
again
i

important-leather-28796

04/03/2019, 7:19 PM
ok
all good now, thanks. I added that process to my rebuild script
f

fierce-dinner-20116

04/03/2019, 7:29 PM
@important-leather-28796 What exactly was the process? Just running
pulumi plugin rm --all
?
f

fierce-dinner-20116

04/03/2019, 7:29 PM
Got it, thanks. I guess I need to add this to CI until this is fixed
i

important-leather-28796

04/03/2019, 7:30 PM
yeah, good point, didn’t think about the same on ci
f

fierce-dinner-20116

04/03/2019, 7:30 PM
It might not matter depending on what your CI tool is
i.e. if it’s installing pulumi from scratch and using a fresh checkout
you might avoid this problem
i

important-leather-28796

04/03/2019, 7:30 PM
just
pulumi plugin rm --all
before yarn may do it, but maybe not if the package is there and this happens on
postinstall
using circleci but haven’t automated yet. just finishing our production by hand today. Should be in cci tomorrow
f

fierce-dinner-20116

04/03/2019, 7:32 PM
tbh, this seems like it could be a p0 if people are using CI/CD with pulumi in production
i

important-leather-28796

04/03/2019, 8:25 PM
somehow the old gcp plugin just appeared again - not sure how, did not re-yarn
Copy code
infrastructure ❯❯❯ pulumi plugin ls                                                                                                                                                                                                                                                          ✘ 255 
NAME        KIND      VERSION  SIZE   INSTALLED       LAST USED
gcp         resource  0.18.2   64 MB  1 hour ago      now
gcp         resource  0.18.0   64 MB  59 minutes ago  now
kubernetes  resource  0.22.0   45 MB  1 hour ago      now
random      resource  0.5.1    27 MB  1 hour ago      now

TOTAL plugin cache size: 199 MB
@stocky-spoon-28903 what would make that happen
s

stocky-spoon-28903

04/03/2019, 8:27 PM
Some kind of reference somewhere. Do you have a tree of custom packages?
i

important-leather-28796

04/03/2019, 8:27 PM
I use a common code package, already sure that is the right npm version
I did restart
tsc
there
not sure how that could impact it
oh, I see the timing, it was shortly after initial use
rebuilding now to re-check installed plugins. I assume only pulumi/gcp npm package installs the gcp plugin?
so plugins were good:
Copy code
~/p/advisorIntake ❯❯❯ pulumi plugin ls
NAME        KIND      VERSION  SIZE   INSTALLED       LAST USED
gcp         resource  0.18.2   64 MB  54 seconds ago  now
kubernetes  resource  0.22.0   45 MB  54 seconds ago  now
random      resource  0.5.1    27 MB  54 seconds ago  now
(after yarn)
then on the first
pulumi up
it installed the older plugin:
Copy code
pulumi plugin ls
NAME        KIND      VERSION  SIZE   INSTALLED       LAST USED
gcp         resource  0.18.2   64 MB  1 minute ago    now
gcp         resource  0.18.0   64 MB  13 seconds ago  7 seconds ago
kubernetes  resource  0.22.0   45 MB  1 minute ago    now
random      resource  0.5.1    27 MB  1 minute ago    now
All declared pulumi packages used in both the stack and shared code:
Copy code
"@pulumi/gcp": "^0.18.2",
"@pulumi/kubernetes": "^0.22.0",
"@pulumi/pulumi": "^0.17.4",
"@pulumi/random": "^0.5.1",
c

cold-coat-35200

04/03/2019, 8:51 PM
The only way I was able to workaround my similar issue(linked in the "Version providers" github issue), is to downgrade the cli to 0.17.1 and remove the unnecessary plugins, the automatic plugin install/update was introduced in 0.17.2
i

important-leather-28796

04/03/2019, 8:53 PM
luke just added a note in the issue, the older version is in the stack export manifest