pub async fn deploy_verify_node(
declared: &Declared,
project: &str,
node: VerifyNode,
master: Option<VerifyMasterSurfaces<'_>>,
master_reader: &dyn TreeReader,
backup: Option<&BackupNode<'_>>,
opts: &VerifyOptions,
) -> Result<DeployVerifyResult, CoreError>Expand description
deploy verify over one node or both (D-12, V6).
Master is deploy_verify unchanged. Backup runs the backup checks
only: the redundancy check (peerConnected and not syncPending), the
mode, the REST read-back of the render, and the backup’s files against the
master’s live copy, read through master_reader (a warn when backup
has no file transport). Both runs the master’s six checks, then the
backup’s. Like deploy_verify it never waits, never resyncs and never
writes: the backup is reached through BackupNode, which has no write
to express, and the master only through reads.
The backup’s read_back follows the master’s standalone rule (R2): a
difference or a missing resource is a warn when the master’s plan
reports that resource same, and red otherwise. The statuses come from
the master verify’s own live-file read (Both), or from the master read
the backup files check makes (Backup). With neither (a backup with no
file transport under --node backup, or a files read that failed), the
backup’s read_back is judged strictly (P10).
The plans are routed as deploy_verify’s (quick 261102-cef, A2), and
the mode-residue warning of a pair comes from the master verify, so
Both carries it once and Backup not at all. Likewise a declared unit’s
mode copy fails the master’s files check (quick 261103-ell, F2) and
leaves the backup checks as they are.
master is required for Master and Both, backup for Backup and
Both; a missing one is CoreError::InvalidInput.