How to retry failed workflow

[leexin]

Hi there,

I'm stuck trying to figure out a way to resume a failed workflow.

Assuming the workflow contains activities A -> B -> C in a sequential manner, and initially the workflow failed because Activity B has a bug that throws Exception.

A (success) -> B (failed) -> C (not run)
Workflow status: failed

After we fix the code bug in B and restart the queue worker, the following code is run:

$workflow = WorkflowStub::load($workflowId);
$workflow->resume();

However, when the workflow resumed at Activity B, the same bug happened even though the bug was fixed and queue worker restarted.
Any ideas how can I resume the workflow from the previously failed activity with fresh source code?

Cheers!

[rmcdaniel]

@leexin Have you taken a look https://laravel-workflow.com/docs/failures-and-recovery?

You shouldn't have to call resume.

[leexin]

Hey @rmcdaniel thanks for the reply. Yes I have read that page.

1. My workflow and activities are set to $tries = 1, so when there's an exception, the workflow will be marked as failed. Since it's already marked as failed, how can I instruct the Workflow to "retry" again?

2. Based on my testing I can't seem to do this "Restart the queue worker to pick up the new code.". After I restart the queue worker, the old code still took effect.

Cheers

[rmcdaniel]

If the workflow status ends with failed that's considered a terminal state. You should avoid letting your workflows become failed. It's okay if an activity "fails" but your workflows should remain in a pending/waiting/etc state so that they can be easily resumed. If you're setting $tries = 1 that is very strict and it will fail the workflow with a single error.

In order to manually retry a failed workflow you will have to do some database editing. You will need to change the workflow status back to something other than failed. That might be all that you need to do to get it working. If not, you might also need to delete the latest entry in the workflow_logs table.

I hope this helps. If not, please let me know where your are stuck or what error you are seeing. Thanks!

[leexin]

Ah, thank you for the explanation. I tried to do the infinite loop method to handle Exception at Workflow and waiting for external signal to continue with the execution.

#[SignalMethod]
public function shouldRetry()
{
    $this->shouldRetry = true;
}

public function execute(array $data = [])
{
    while (true) {
        try {
            yield ActivityStub::make(ActivityOne::class);
            $step_1_result = yield ActivityStub::make(ActivityTwo::class, $data);   // assuming this activity threw Exception
            $step_2_result = yield ActivityStub::make(ActivityThree::class);
            $step_3_result = yield ActivityStub::make(ActivityFour::class, $data);

            break;
        } catch (\Throwable $e) {
            // pause until signal receive
            yield WorkflowStub::await(fn () => $this->shouldRetry);
            $this->shouldRetry = false;
        }
    }
}

But the failed ActivityTwo does not automatically retry after the signal shouldRetry is received. I need to manually delete the row containing Exception from workflow_logs table for the Workflow to reattempt the failed ActivityTwo. Do note that my Activity is set to $tries = 1, and workflow is set to $tries = 0 now.

Is this the correct retry flow? or am I missing something.

[rmcdaniel]

I think it's a bug. I will investigate it and report back.

[rmcdaniel]

@leexin Yes, it was a bug. Your code should work if you update to version 1.0.39. Please let me know if it does not fix it.