https://pulumi.com logo
Docs
Join the conversationJoin Slack
Channels
announcements
automation-api
aws
azure
blog-posts
built-with-pulumi
cloudengineering
cloudengineering-support
content-share
contribex
contribute
docs
dotnet
finops
general
getting-started
gitlab
golang
google-cloud
hackathon-03-19-2020
hacktoberfest
install
java
jobs
kubernetes
learn-pulumi-events
linen
localstack
multi-language-hackathon
office-hours
oracle-cloud-infrastructure
plugin-framework
pulumi-cdk
pulumi-crosscode
pulumi-deployments
pulumi-kubernetes-operator
pulumi-service
pulumiverse
python
registry
status
testingtesting123
testingtesting321
typescript
welcome
workshops
yaml
Powered by Linen
general
  • s

    stocky-spoon-28903

    08/07/2018, 6:15 PM
    But ultimately the number would need to be stringified to get down to the provider
  • s

    stocky-spoon-28903

    08/07/2018, 6:15 PM
    It’s a minor usability detail in the scheme of things but one that just cost me 15 minutes of looking at a type error
  • m

    microscopic-florist-22719

    08/07/2018, 6:16 PM
    Ah, interesting. There's no support for this today, but I can't imagine that it would be too terribly difficult to add.
  • s

    stocky-spoon-28903

    08/07/2018, 6:18 PM
    Ultimately I guess the different SDKs would have to treat it differently - I don’t know of a way to represent it in Go for example
  • m

    microscopic-florist-22719

    08/07/2018, 6:21 PM
    That's right.
  • s

    stocky-spoon-28903

    08/07/2018, 6:23 PM
    Actually breaking out the healthcheck argument into a custom type would make the error a lot better. I’ll do that later today.
    w
    • 2
    • 1
  • s

    stocky-spoon-28903

    08/07/2018, 6:23 PM
    As it is I got an error about the entire arguments struct
  • p

    purple-appointment-43233

    08/07/2018, 6:40 PM
    Hey, so I have an (I think) interesting problem: I use Pulumi to construct a SpotFleetRequest, and if I change the contents of the request, pulumi will destroy the old one and create a new one. This will leave a period of time where I have zero spot instances running. I’d like to have some way to specify an orderly transition, ideally have the old spot instances drain out slowly while the new ones take their place.
  • w

    white-balloon-205

    08/07/2018, 6:59 PM
    Pulumi will always create a replacement before it deletes the old resource. So if you are directly managing resources like Instances or Functions or Containers, you will get this orderly transition. However, I'm not off the top of my head sure what guarantees Spot Fleet makes around availability of instances at the time the fleet creation "succeeds". If it has not been designed to support this, it may require some extra work to handle this via Pulumi (or really via any system that coordinates Spot Fleets).
  • w

    white-balloon-205

    08/07/2018, 7:01 PM
    Ahh - it looks like there is a
    waitForFullfillment
    which can be used to not register the creation as successful until the request has completed. That should ensure that the delete doesn't happen until the create suceeded in bringing up the new instances. https://pulumi.io/reference/pkg/nodejs/@pulumi/aws/ec2/#property-waitforfulfillment
    p
    • 2
    • 4
  • s

    stocky-spoon-28903

    08/07/2018, 7:02 PM
    Something else I’ve seen people do in this specific case is have lambda manage effectively blue-green deployment of spot requests, and have {Terraform,Pulumi} adjust the lambda instead
  • s

    stocky-spoon-28903

    08/07/2018, 7:13 PM
    Huh… Pulumi just handled my moving a bunch of resources between various components.
  • s

    stocky-spoon-28903

    08/07/2018, 7:13 PM
    Without having to destroy and recreate them.
    w
    • 2
    • 2
  • s

    stocky-spoon-28903

    08/07/2018, 8:59 PM
    Fun little component that creates a bastion host in a fairly customizable manner: https://gist.github.com/jen20/730e392f8d8361f02f494b294f62d66c
    💥 1
  • s

    stocky-spoon-28903

    08/07/2018, 9:38 PM
    Just ran into a new (to me) panic that doesn’t really give any description of where things are: https://gist.github.com/jen20/8cd9ce3b7687a22ad2a0f7f0331601dd
  • s

    stocky-spoon-28903

    08/07/2018, 9:41 PM
    Working to find a minimal repro at the moment
    m
    • 2
    • 8
  • a

    adamant-restaurant-73893

    08/07/2018, 10:18 PM
    For those of you who signed up for tshirts, looks like the inventory is finally ready and they're shipping. 🎉 and thanks for your patience. If you still want a tshirt, just write some nice things here and we'll send one to you too. http://info.pulumi.com/community/give-me-a-tshirt
  • a

    adamant-restaurant-73893

    08/07/2018, 10:19 PM
    Also, if you're speaking at local meetups or similar, let me know, and I'll hook you up with stickers.
  • s

    strong-helicopter-19655

    08/08/2018, 1:19 PM
    Are there size/fit options?
    a
    • 2
    • 1
  • a

    adventurous-jordan-10043

    08/08/2018, 2:35 PM
    yes 🙂
  • s

    stocky-spoon-28903

    08/08/2018, 2:56 PM
    Is there a complelling reason not to set
    --parallel
    to a higher default at the moment?
    c
    i
    • 3
    • 3
  • s

    stocky-spoon-28903

    08/08/2018, 3:13 PM
    Although interestingly it doesn’t make the difference I would have expected set to 5
  • i

    incalculable-sundown-82514

    08/08/2018, 3:18 PM
    That’s interesting - what did you see?
  • s

    stocky-spoon-28903

    08/08/2018, 3:19 PM
    Basically serial operation on update, preview seemed faster
  • s

    stocky-spoon-28903

    08/08/2018, 3:20 PM
    Not sure if the structure of my program is making that happen though
  • i

    incalculable-sundown-82514

    08/08/2018, 3:22 PM
    ah, OK. You can use
    pulumi stack graph
    to see your program’s dependency graph - if we’re failing to parallelize independent nodes in the graph, that would be a bug for us
  • s

    stocky-spoon-28903

    08/08/2018, 3:57 PM
    I’ll try to get a better idea of that shortly. I’m running the latest released version rather than from source now
  • w

    white-balloon-205

    08/08/2018, 3:58 PM
    @stocky-spoon-28903 Note that - although there has been a
    —parallel
    flag since launch, it has not actually been fully wired through to run deployments in parallel. We just landed the first support for full deployment parallelism in master on Monday, and expect this to be enabled in the next release.
    s
    • 2
    • 2
  • p

    purple-appointment-43233

    08/08/2018, 5:45 PM
    Hey folks, so I did a
    pulumi update
    and it hung forever, so I double-ctrl-c and now I’m in an inconsistent state.
  • p

    purple-appointment-43233

    08/08/2018, 5:45 PM
    What do I do? The provided link has no information
    c
    w
    • 3
    • 14
Powered by Linen
Title
p

purple-appointment-43233

08/08/2018, 5:45 PM
What do I do? The provided link has no information
c

colossal-beach-47527

08/08/2018, 5:46 PM
Sorry to hear that @purple-appointment-43233, what was the link that you got? We’ll be sure to fix that ASAP. As far as how to handle this state, you just need to do the following:
p

purple-appointment-43233

08/08/2018, 5:46 PM
error: [409] the stack’s resources are in an unknown state due to an interrupted update; please export the stack, repair any inconsistencies, and import the result for more information about this error, see https://goo.gl/hCQnhn
c

colossal-beach-47527

08/08/2018, 5:47 PM
Export your Stack’s current checkpoint file. This essentially dumps the current known state of your cloud resources.
pulumi stack export > checkpoint.json
Next, you can import that checkpoint and overwrite the set of resources known on the stack.
pulumi stack import < checkpoint.json
At that point the “dirty” bit should be cleared. Let me know if you have any other trouble!
p

purple-appointment-43233

08/08/2018, 5:49 PM
Looks like that cleared it. But I’m still running into a problem where a very simple AWS deploy appears to be hanging forever making a spotfleetrequest
c

colossal-beach-47527

08/08/2018, 5:50 PM
As far as insight into that, I don’t have any ideas off the top of my head. Some resources do take a very long time to provision. Any ideas @white-balloon-205?
p

purple-appointment-43233

08/08/2018, 5:50 PM
I changed my spotfleetrequest spec to use a new ami and to last for 10 years, and it just hangs attempting to deploy
Weird, after about 5 minutes, AWS barfed saying that the ami doesn’t exist. Which is weird because it’s the latest AWS ECS-optimized ami listed on their website…
Well, retrying until it worked…worked. ¯\_(ツ)_/¯
Thanks for your help
w

white-balloon-205

08/08/2018, 6:10 PM
Note that AWS AMI IDs are per-region, so you'll need to make sure you pick the AMI ID that matches the region you are deploying into. You can also use Paramater Store to get this via https://aws.amazon.com/about-aws/whats-new/2018/04/amazon-ecs-provides-ecs-optimized-ami-metadata-via-ssm-parameter/. See https://github.com/pulumi/pulumi-aws/blob/dc0278d889de6d868c7497a473c32b709aa13521/examples/multiple-regions/index.ts#L12 for an example.
As for why this takes so long - that seems unexpected - AWS should normally fail these quickly. Possible that SpotFeetRequest doesn't validate these things early until Spot fulfills the request.
p

purple-appointment-43233

08/08/2018, 6:22 PM
Sweet! It’s getting meta in here. BTW, the example you gave me is no longer syntactically correct, getParameter doesn’t appear to take a second arg, but I did get it working. 🙂
w

white-balloon-205

08/08/2018, 6:42 PM
Yeah - sorry - that example is part of an in-progress PR which also adds the ability to pass that second parameter - but your question reminded me of it as we were discussing this on the PR just yesterday.
View count: 1