Skip to contents

Get running processes

Source: R/processes.R
get_processes.Rd

Returns the processes that Connect has started for content and that have not yet finished, with their most recent CPU and memory use. This is the data shown on the System > Processes page of the Connect dashboard. It requires administrator privileges.

Usage

get_processes(src, content_guid = NULL, owner_guid = NULL, hostname = NULL)

Arguments

src

A Connect R6 client object.

content_guid

Optional. Return only processes for the content item with this GUID. It is an error if no content item has this GUID.

owner_guid

Optional. Return only processes for content owned by the user with this GUID. It is an error if no user has this GUID.

hostname

Optional. Return only processes running on the Connect node with this hostname.

Value

A tibble with one row per process and the following columns:

  • content_guid: The GUID of the content item the process belongs to. NA when the content item no longer exists.

  • content_name: The title of the content item, or its name when it has no title.

  • content_url: The URL of the content item, including its custom URL if it has one.

  • run_as: The Unix user the process runs as.

  • type: The kind of work the process does, such as run_app or build_report.

  • pid: The process identifier on the Connect host. NA when Connect has not recorded one, such as for off-host execution.

  • hostname: The name of the Connect node that started the process.

  • job_key: The key of the job for this process.

  • image: The container image the process runs in, when using off-host execution.

  • start_time: When the process started.

  • heartbeat: When the last CPU and memory measurement was taken.

  • cpu_current: The fraction of one CPU the process used over the last measurement interval.

  • ram: The resident memory of the process at the last measurement, as fs::fs_bytes.

Details

The results cover every node in a Highly Available cluster and include processes that run with off-host execution (e.g. Kubernetes).

Note that:

  • cpu_current, ram, and heartbeat hold the last measurement each process reported, so they can be a few seconds old. They are NA until a process reports its first measurement.

  • A process that exited recently can still appear until Connect finishes cleaning up after it.

On Connect versions without the GET /v1/system/processes endpoint, this falls back to the older, unversioned GET /metrics/procs endpoint. The results are converted to the same columns, and the filters are applied client-side.

When more than one filter is given, a process must match all of them.

Examples

if (FALSE) { # \dontrun{
client <- connect()

get_processes(client)

# Only the processes for one content item
get_processes(client, content_guid = "6f300623-1e0c-48e6-a473-ddf630c0c0c3")
} # }