Class: ChronoForge::Dashboard::ActionsController

Inherits:
BaseController
  • Object
show all
Defined in:
app/controllers/chrono_forge/dashboard/actions_controller.rb

Instance Method Summary collapse

Instance Method Details

#bulk_reap ⇒ Object

Reap every stranded workflow (running with a stale lock) in the background — the bulk form of #reap, delegating to Workflow.reap_stalled. The count is taken up front for the flash; the job does the sweep.



44
45
46
47
48
49
50
 
# File 'app/controllers/chrono_forge/dashboard/actions_controller.rb', line 44

def bulk_reap
  cutoff = ChronoForge.config.reap_stale_after.ago
  n = ChronoForge::Workflow.where(state: ChronoForge::Workflow.states[:running]).where(locked_at: ...cutoff).count
  return redirect_to(stranded_index_path, notice: "No stranded workflows to reap.", status: :see_other) if n.zero?
  BulkReapJob.perform_later
  redirect_to stranded_index_path, notice: "Reaping #{n} stranded workflow(s) in the background.", status: :see_other
end

#bulk_retry ⇒ Object

Retry every blocked (failed/stalled) workflow. The fan-out runs in a background job so the request returns fast even with a huge backlog; the count is taken up front for the flash (BulkRetryJob does the enqueueing).



55
56
57
58
59
60
 
# File 'app/controllers/chrono_forge/dashboard/actions_controller.rb', line 55

def bulk_retry
  n = BulkRetryJob.retryable.count
  return redirect_to(workflows_path, notice: "No blocked workflows to retry.", status: :see_other) if n.zero?
  BulkRetryJob.perform_later
  redirect_to workflows_path, notice: "Retrying #{n} blocked workflow(s) in the background.", status: :see_other
end

#bulk_retry_branch ⇒ Object

Retry every blocked (failed/stalled) child of one branch, in the background.



63
64
65
66
67
68
69
70
71
 
# File 'app/controllers/chrono_forge/dashboard/actions_controller.rb', line 63

def bulk_retry_branch
  parent = ChronoForge::Workflow.find(params[:workflow_id])
  branch_log = parent.execution_logs.find(params[:id])
  n = BulkRetryJob.retryable(branch_log).count
  redirect = ->(msg) { redirect_to workflow_branch_path(parent, branch_log), notice: msg, status: :see_other }
  return redirect.call("No blocked child workflows to retry.") if n.zero?
  BulkRetryJob.perform_later(branch_log.id)
  redirect.call("Retrying #{n} child workflow(s) in the background.")
end

#reap ⇒ Object

Recover a workflow stranded in :running — the single-workflow form of the stalled-workflow reaper (Workflow.reap_stalled). Re-enqueues it so the executor's acquire_lock steals the stale lock and replays completed steps as no-ops. This is the right fix for an overdue/stuck run: unlike Force unlock (which idles the row and leaves nothing to wake it), reap actually re-drives the workflow. Safe and idempotent — a duplicate just loses the acquire_lock race and no-ops.



25
26
27
28
29
 
# File 'app/controllers/chrono_forge/dashboard/actions_controller.rb', line 25

def reap
  return redirect_to(workflow_path(workflow), alert: "Only running workflows can be reaped.", status: :see_other) unless workflow.running?
  workflow.job_klass.perform_later(workflow.key, **workflow.kwargs.symbolize_keys)
  redirect_to workflow_path(workflow), notice: "Reaped #{workflow.key} — re-enqueued to steal its stale lock and replay.", status: :see_other
end

#resume ⇒ Object

Re-enqueue an idle (parked) workflow so the executor picks it up again. This is the recovery for a dropped poll/wake — a wait_until or merge whose poller job was lost, or a continue_if whose event has since arrived: the replay re-checks the condition and re-arms the poll if still unmet.



35
36
37
38
39
 
# File 'app/controllers/chrono_forge/dashboard/actions_controller.rb', line 35

def resume
  return redirect_to(workflow_path(workflow), alert: "Only idle workflows can be resumed.", status: :see_other) unless workflow.idle?
  workflow.job_klass.perform_later(workflow.key)
  redirect_to workflow_path(workflow), notice: "Re-enqueued #{workflow.key}.", status: :see_other
end

#retry ⇒ Object 



8
9
10
11
 
# File 'app/controllers/chrono_forge/dashboard/actions_controller.rb', line 8

def retry
  workflow.retry_later
  redirect_to workflow_path(workflow), notice: "Re-enqueued #{workflow.key}.", status: :see_other
end

#unlock ⇒ Object 



13
14
15
16
 
# File 'app/controllers/chrono_forge/dashboard/actions_controller.rb', line 13

def unlock
  workflow.update!(locked_at: nil, locked_by: nil, state: :idle)
  redirect_to workflow_path(workflow), notice: "Unlocked #{workflow.key}.", status: :see_other
end