jobs à la cron using the payload jobs queue

linobino1last year

19

export default buildConfig({
    jobs: {
    tasks: [
      {
        slug: 'purgeDatabase',
        inputSchema: [],
        outputSchema: [],
        handler: async ({ req: { payload } }) => {
          payload.logger.info('purging the database...')
          // ... do something
          return {
            output: undefined,
          }
        },
      },
    ],
    autoRun: [
      {
        queue: 'nightly',
        cron: '0 0 * * *',
      },
    ],
  },
  onInit: (payload) => {
    payload.jobs.queue({ task: 'purgeDatabase', queue: 'nightly', input: {} })
  },
  // ... more config

• 

  zed0547last year

  Are you on Vercel? If so, see the bottom of this section:

  https://payloadcms.com/docs/jobs-queue/queues#cron-jobs
• 

  linobino1last year

  I'll be on a VPS but right now in development only

• 

  zed0547last year

  Your dev machine has the cron service running?

• 

  linobino1last year

  I'm literally running above code in a blank payload in stall with

  pnpm dev

  , not more...

  
  
  

  is this not supposed to work in dev?

  
  
  

  autoRun is being triggered

• 

  zed0547last year

  It depends I imagine if you actually have a cron service on your machine

• 

  linobino1last year

  hmm but why would it work once then?

  
  
  

  it triggers the task on the correct time, but only once

  
  
  

  [16:32:00] INFO: Running 1 jobs.
  [16:32:00] INFO: purging the database...

• 

  zed0547last year

  Yeah it's weird it gets triggered once at the correct time

  
  
  

  Not sure to be honest

• 

  linobino1last year

  thanks anyways

  
  
  

  it feels a bit to me like there was something missing for tasks to repeat... like whenever the task runs, it gets removed from the queue and that's it

  
  
  

  something like

  repeat: true
• 

  aaronksaunderslast year

  @1024331309486719056

  can i suggest just requeuing the task/job on success?

• 

  linobino1last year

  you may;)

  
  
  

  but actually, the

  onSuccess

  handler does not have any props, so the payload object is not available

  
  
  

    jobs: {
      tasks: [
        {
          slug: 'purgeDatabase',
          retries: Infinity,
          handler: async ({ req: { payload } }) => {
            // ... do something
  
            // renew the task
            payload.jobs.queue({ task: 'purgeDatabase', queue: 'nightly', input: {} })
            return {
              output: undefined,
            }
          },
        },
      ],
      autoRun: [
        {
          queue: 'nightly',
          limit: Infinity,
          cron: '* * * * *',
        },
      ],
    },

  
  
  

  this is what I will use then

• 

  aaronksaunderslast year

  i am curious, so adding limit infinity worked? interesting that in the on success to dont get any props added...

• 

  linobino1last year

  only time will tell:)

• 

  aaronksaunderslast year

  i see u just added the re-queuing to the original task 👍🏿

• 

  nomad.devlast year

  Hey, I was leading to the same way of doing but it seems a bit weird to me to have to do this workaround. Did you find any other / cleaner way to do that?

• 

  linobino1last year

  I haven't, I also did not deploy this thing yet

• 

  nomad.devlast year

  Depending on the usecase, you might wan't to include the re-queue in a

  try {} catch (e) {} finally { [enqueue] }

  to avoid loosing the re-enqueuing in case of errors.

• 

  linobino1last year

  that's what I did, thanks.

  
  
  

  Queueing the job in

  onInit()

  as sketched above will duplicate the job on each startup. I queued my jobs in a migration script now.

  
  
  

  I wanted to allow manually triggering the job too via an endpoint, so I added an input parameter

  
  
  

  inputSchema: [
      {
        name: 'repeat',
        type: 'checkbox',
        defaultValue: false,
      },
    ],

  
  
  

  depending on which the job will be requeued in the

  finally { }

  clause.

• 

  nomad.devlast year

  Smart, thanks !

• 

  linobino1last year

  this did not prove to be reliable... it might work for a few days and then it just stops running the job.