Skip to contents

Get a full DAG overview of all flows and their health

Source: R/deps.R
pl_get_dag.Rd

Returns a tibble (data.frame) of ALL flows with both their raw (most recent log) status and effective (cascade-aware) status. A flow is considered "poisoned" if any upstream dependency has a more recent ERROR/FATAL log than the flow's own last SUCCESS log — meaning the flow ran successfully before the upstream broke.

Usage

pl_get_dag(conn, since = NULL)

Arguments

conn

A connection object from pl_connect().

since

Optional POSIXct or ISO 8601 string. If provided, only considers log entries created after this timestamp.

Value

A data.frame with columns:

  • flow (character): flow name

  • type (character): flow type

  • schedule (character): schedule string

  • raw_status (character): most recent log status (NA if never logged)

  • raw_status_time (character): timestamp of most recent log

  • effective_status (character): "SUCCESS", "ERROR", "FATAL", "POISONED", or NA if never logged

  • poisoned_by (list-column): character vector of upstream flow names that caused poisoning

  • depends_on (list-column): character vector of direct upstream flow names

  • is_root (logical): TRUE if the flow has no upstream dependencies

Details

Poisoning rule: A downstream flow is poisoned if:

  1. It has at least one upstream flow with an ERROR or FATAL log.

  2. The downstream's last SUCCESS log is MORE RECENT than that upstream failure. (i.e., the downstream ran successfully after the upstream had already broken.)

  3. The downstream itself is not already ERROR or FATAL.

If the downstream has never logged, effective_status is NA (unknown), regardless of upstream status.

Examples

if (FALSE) { # \dontrun{
conn <- pl_connect()
dag <- pl_get_dag(conn)
dag
} # }