You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Migrating from Bucket to BucketV2 in Pulumi without losing data is becoming more difficult than it used to be in 2022 due to continued evolution and schema drift. Consider building support to make it more seamless to support users that need to perform the migration.
Technically there are a couple of things that could be done:
To overcome schema drift some code could be built to parse Bucket states into BucketV2 schema and automatically migrate them; this seems to be something that worked in 2022 but is not entirely seamless now as the schema of BucketV2 evolved separately. This would help users to simply rename Bucket to BucketV2 without moving deprecated inputs like serverSideEncryptionConfiguration to side-by-side resources and do a pulumi up.
Actually supporting moving deprecated inputs to side-by-side resources is a lot more involved since by default Pulumi does not correlate the states properly and will not send the prior Bucket's state to the new BucketServerSideEncryptionConfiguration resource. Document migrating aws.s3.Bucket to aws.s3.BucketV2 registry#5444 outlines manual steps but with some work it may be possible to partially automate them in a tool, possibly involving some form of automated import or refresh.
If nothing is done to simplify this scenario, migrating without losing data is still possible. We're finalizing a guide on how to do this.
Your feedback helps Pulumi team prioritize, please upvote this issue if you are affected by the migration and would like to see improved automated tooling support.
Thank you for raising this, highly appreciated. I favor stability and the possibility to migrate forward over almost everything else when it comes to infrastructure-as-code tools.
I wouldn't mind having to change the code a bit, e.g. moving from lifecycleRules to their own objects, as long as Pulumi picks up that the bucket is in the desired state already.
Migrating from Bucket to BucketV2 in Pulumi without losing data is becoming more difficult than it used to be in 2022 due to continued evolution and schema drift. Consider building support to make it more seamless to support users that need to perform the migration.
Technically there are a couple of things that could be done:
To overcome schema drift some code could be built to parse Bucket states into BucketV2 schema and automatically migrate them; this seems to be something that worked in 2022 but is not entirely seamless now as the schema of BucketV2 evolved separately. This would help users to simply rename Bucket to BucketV2 without moving deprecated inputs like serverSideEncryptionConfiguration to side-by-side resources and do a
pulumi up
.Actually supporting moving deprecated inputs to side-by-side resources is a lot more involved since by default Pulumi does not correlate the states properly and will not send the prior Bucket's state to the new BucketServerSideEncryptionConfiguration resource. Document migrating aws.s3.Bucket to aws.s3.BucketV2 registry#5444 outlines manual steps but with some work it may be possible to partially automate them in a tool, possibly involving some form of automated import or refresh.
If nothing is done to simplify this scenario, migrating without losing data is still possible. We're finalizing a guide on how to do this.
Your feedback helps Pulumi team prioritize, please upvote this issue if you are affected by the migration and would like to see improved automated tooling support.
References:
The text was updated successfully, but these errors were encountered: