Remove pageserver from deploy script

.github/ansible/prod.us-west-2.hosts.yaml

@@ -43,8 +43,6 @@ storage:
           ansible_host: i-00c3844beb9ad1c6b
         pageserver-4.us-west-2.aws.neon.tech:
           ansible_host: i-013263dd1c239adcc
-        pageserver-5.us-west-2.aws.neon.tech:
-          ansible_host: i-00ca6417c7bf96820
         pageserver-6.us-west-2.aws.neon.tech:
           ansible_host: i-01cdf7d2bc1433b6a
         pageserver-7.us-west-2.aws.neon.tech:

compute_tools/src/compute.rs

@@ -338,30 +338,27 @@ impl ComputeNode {
         // In this case we need to connect with old `zenith_admin` name
         // and create new user. We cannot simply rename connected user,
         // but we can create a new one and grant it all privileges.
-        let mut client = {
-            let _span = tracing::info_span!("connect").entered();
-            match Client::connect(self.connstr.as_str(), NoTls) {
-                Err(e) => {
-                    info!(
-                        "cannot connect to postgres: {}, retrying with `zenith_admin` username",
-                        e
-                    );
-                    let mut zenith_admin_connstr = self.connstr.clone();
+        let mut client = match Client::connect(self.connstr.as_str(), NoTls) {
+            Err(e) => {
+                info!(
+                    "cannot connect to postgres: {}, retrying with `zenith_admin` username",
+                    e
+                );
+                let mut zenith_admin_connstr = self.connstr.clone();
 
-                    zenith_admin_connstr
-                        .set_username("zenith_admin")
-                        .map_err(|_| anyhow::anyhow!("invalid connstr"))?;
+                zenith_admin_connstr
+                    .set_username("zenith_admin")
+                    .map_err(|_| anyhow::anyhow!("invalid connstr"))?;
 
-                    let mut client = Client::connect(zenith_admin_connstr.as_str(), NoTls)?;
-                    client.simple_query("CREATE USER cloud_admin WITH SUPERUSER")?;
-                    client.simple_query("GRANT zenith_admin TO cloud_admin")?;
-                    drop(client);
+                let mut client = Client::connect(zenith_admin_connstr.as_str(), NoTls)?;
+                client.simple_query("CREATE USER cloud_admin WITH SUPERUSER")?;
+                client.simple_query("GRANT zenith_admin TO cloud_admin")?;
+                drop(client);
 
-                    // reconnect with connsting with expected name
-                    Client::connect(self.connstr.as_str(), NoTls)?
-                }
-                Ok(client) => client,
+                // reconnect with connsting with expected name
+                Client::connect(self.connstr.as_str(), NoTls)?
             }
+            Ok(client) => client,
         };
 
         // Proceed with post-startup configuration. Note, that order of operations is important.

pageserver/src/http/openapi_spec.yml

@@ -340,29 +340,7 @@ paths:
           format: hex
 
     post:
-      description: |
-        Schedules attach operation to happen in the background for the given tenant.
-        As soon as the caller sends this request, it must assume the pageserver
-        starts writing to the tenant's S3 state unless it receives one of the
-        distinguished errors below that state otherwise.
-
-        The method to identify whether a request has arrived at the pageserver, and
-        whether it has succeeded, is to poll for the tenant status to reach "Active"
-        or "Broken" state. These values are currently not explicitly documented in
-        the API spec.
-        Polling for `has_in_progress_downloads == false` is INCORRECT because that
-        value can turn `false` during shutdown while the Attach operation is still
-        unfinished.
-
-        There is no way to cancel an in-flight request.
-
-        If a client receives a not-distinguished response, e.g., a network timeout,
-        it MUST retry the /attach request and poll again for tenant status.
-
-        In any case, it must
-        * NOT ASSUME that the /attach request has been lost in the network,
-        * NOT ASSUME that the request has been lost based on a subsequent
-          tenant status request returning 404. (The request may still be in flight!)
+      description: Schedules attach operation to happen in the background for given tenant
       responses:
         "202":
           description: Tenant attaching scheduled