fix: deny branching, starting compute from not yet uploaded timelines (#5484)

Part of #5172. First commits show that we used to allow starting up a
compute or creating a branch off a not yet uploaded timeline. This PR
moves activation of a timeline to happen **after** initial layer file(s)
(if any) and `index_part.json` have been uploaded. Simply moving
activation to be *after* downloads have finished works because we now
spawn a task per http request handler.

Current behaviour of uploading on the timelines on next startup is kept,
to be removed later as part of #5172.

Adds:
- `NeonCli.map_branch` and corresponding `neon_local` implementation:
allow creating computes for timelines managed via pageserver http
client/api
- possibly duplicate tests (I did not want to search for, will cleanup
in a follow-up if these duplicated)

Changes:
- make `wait_until_tenant_state` return immediatedly on `Broken` and not
wait more
This commit is contained in:
Joonas Koivunen
2023-10-09 17:03:38 +03:00
committed by GitHub
parent 010b4d0d5c
commit 4772cd6c93
7 changed files with 345 additions and 10 deletions
+3
View File
@@ -396,6 +396,9 @@ async fn timeline_create_handler(
format!("{err:#}")
))
}
Err(e @ tenant::CreateTimelineError::AncestorNotActive) => {
json_response(StatusCode::SERVICE_UNAVAILABLE, HttpErrorBody::from_msg(e.to_string()))
}
Err(tenant::CreateTimelineError::Other(err)) => Err(ApiError::InternalServerError(err)),
}
}
+10 -2
View File
@@ -406,6 +406,8 @@ pub enum CreateTimelineError {
AlreadyExists,
#[error(transparent)]
AncestorLsn(anyhow::Error),
#[error("ancestor timeline is not active")]
AncestorNotActive,
#[error(transparent)]
Other(#[from] anyhow::Error),
}
@@ -1587,6 +1589,12 @@ impl Tenant {
.get_timeline(ancestor_timeline_id, false)
.context("Cannot branch off the timeline that's not present in pageserver")?;
// instead of waiting around, just deny the request because ancestor is not yet
// ready for other purposes either.
if !ancestor_timeline.is_active() {
return Err(CreateTimelineError::AncestorNotActive);
}
if let Some(lsn) = ancestor_start_lsn.as_mut() {
*lsn = lsn.align();
@@ -1619,8 +1627,6 @@ impl Tenant {
}
};
loaded_timeline.activate(broker_client, None, ctx);
if let Some(remote_client) = loaded_timeline.remote_client.as_ref() {
// Wait for the upload of the 'index_part.json` file to finish, so that when we return
// Ok, the timeline is durable in remote storage.
@@ -1632,6 +1638,8 @@ impl Tenant {
})?;
}
loaded_timeline.activate(broker_client, None, ctx);
Ok(loaded_timeline)
}
@@ -31,6 +31,7 @@ pub(super) async fn upload_index_part<'a>(
fail_point!("before-upload-index", |_| {
bail!("failpoint before-upload-index")
});
pausable_failpoint!("before-upload-index-pausable");
let index_part_bytes =
serde_json::to_vec(&index_part).context("serialize index part file into bytes")?;