Allow redeployment of grafted subgraph even when graft_base is not available #4695
Conversation
fd791dd to
8d2ca4d
Compare
8d2ca4d to
6ea0edf
Compare
…base is not available
6ea0edf to
a54409d
Compare
| } | ||
|
|
||
| pub(crate) fn exists(&self, id: Arc<Site>) -> Result<bool, StoreError> { | ||
| let conn = self.get_conn()?; |
There was a problem hiding this comment.
Lets use with_conn here to make this async.
There was a problem hiding this comment.
@leoyvens i tried that initially wouldn't that mean i would have to make all the callers also async function to await on it?
There was a problem hiding this comment.
I was unsure if i should do that, let me know if its ok.
There was a problem hiding this comment.
Are there any non-async callers? If there are then we'd need to keep this one and add an async version.
There was a problem hiding this comment.
fn create_deployment_internal is not async but its callers can be made async ( cause callers all the way is async ). Should i make them all async in that case?
There was a problem hiding this comment.
I see now that this is a bigger problem since fn create_deployment_internal is doing many sync DB operations, but is called from an async context. Ok we can leave this exists as sync for now, and later find a solution for the whole function.
store/postgres/src/subgraph_store.rs
Outdated
| layout.site.schema_version | ||
| } else { | ||
| DeploymentSchemaVersion::LATEST | ||
| }; |
There was a problem hiding this comment.
I think we could avoid the need for a schema_version variable here by having fn allocate_site take a graft_base parameter instead of schema_version, and do the logic there.
fn allocate_site already checks for existence, and if it needs to search for the graft base I believe it could use fn find_active_site.
There was a problem hiding this comment.
got it, will make that change
core/src/subgraph/registrar.rs
Outdated
| .map_err(SubgraphRegistrarError::ResolveError) | ||
| .await?; | ||
|
|
||
| let exists = store.is_deployed(&deployment)?; |
There was a problem hiding this comment.
There is an fn graft_pending which we perhaps should additionally check, because it checks if the subgraph has actually finished the graft process. It would be nice if we were able to use that here.
dde11bc to
8ccd608
Compare
|
Integration tests, failing.. checking it out. |
8ccd608 to
5062fe8
Compare
leoyvens
left a comment
There was a problem hiding this comment.
Looks great! Thanks for bearing with the reviews
| Ok(graft_pending) => graft_pending, | ||
| Err(StoreError::DeploymentNotFound(_)) => true, | ||
| Err(e) => return Err(SubgraphRegistrarError::StoreError(e)), | ||
| }; |
There was a problem hiding this comment.
Would be great to have comments here.
store/postgres/src/primary.rs
Outdated
| /// Create a site for a brand new deployment. | ||
| /// If it already exists, return the existing site. | ||
| /// and a boolean indicating whether a new site was created. | ||
| /// false means the site already existed. |
There was a problem hiding this comment.
Nice comment, just need to check ponctuation and capitalization.
There was a problem hiding this comment.
ah thanks for catching that. I'll fix that
f999321 to
c73fc91
Compare
core/src/subgraph/registrar.rs
Outdated
| // Determine if the subgraph should be validated. | ||
| // Validate the subgraph if there is a pending graft, indicating the presence of a graft base. | ||
| // If the subgraph is new (DeploymentNotFound), it should also be validated. | ||
| // If the subgraph already exists and there is no pending graft, validation is not required. |
There was a problem hiding this comment.
The wording here is not very precise, this boolean is not deciding if the whole subgraph should be validated, but if the graft base should be validated.
There was a problem hiding this comment.
// Determine if the graft_base should be validated.
// Validate the graft_base if there is a pending graft, ensuring its presence.
// If the subgraph is new (indicated by DeploymentNotFound), the graft_base should be validated.
// If the subgraph already exists and there is no pending graft, graft_base validation is not required.
@leoyvens how does this look?
c73fc91 to
f275665
Compare
Closes #4534