How reliable are Tasks?

Hi all,

I have built a safety app that allows users to notify ‘an authority’ they are heading out , where they are going and when they are due back and once back the user closes down their journey. If they are not back by their due back time.

i have multiple pairs of tasks that run during the day and email anyone who has indicated they have not returned and it is past their due back time. The second task in the pair runs 2 mins later and changes a record to notified, so when the pair of tasks run again they don’t get a second email.

This has been running well for a week, but today the first tas has run 20 mins late, but the second task ran on time, which meant no one got emailed as the second task changed the flag to notified. This resulted in multiple escalations of the emergency services for non returned people, who just hadn’t been reminded to close down their journey logs.

Can tasks be relied upon with Knack to run on time? Is this delay a one off or a regular occurrence?

It seems crazy if there is known unreliability of tasks to run on time, that you cannot carry out multi-stage tasks as a change state after email sent is pretty much a given for any task-based automation.

Hi Andrew,

From experience and discussions with the Knack Support Team, I know that there can be occasions when Tasks are delayed simply due to load / demand. It could be worth raising a support ticket to dive into your specific situation as having multiple pairs of tasks run only 2 mins apart could be quite tight.

Over the past few years I have never had a issue with tasks being unreliable and have many doing updates and email notifications. I think this is likely just to do with volume of requests and scheduling.

Hope this helps.

1 Like

Hi Andrew,
Like Carl mentioned Tasks can sometimes fall behind. They can run into what we call concurrency problems as there is a limit of four jobs (including tasks) running at a time per app, as well as 14 per server.
One of the first things we recommend when people run into tasks backing up over each other is to stagger their times more like 15 minutes than 2 minutes apart. Do that and then please let us know at if something is still not working correctly in your task workflow.

You may also want to check out this other related post about tasks with dependencies on other tasks like yours. Linked tasks (task dependency)
It includes some good information from a Knack Expert and may help you know more about how our task system handles tasks.


Thanks all. I’ve had a good chat with support and understand the situation.

I cannot space out the tasks more than a few minutes, 5 mins max due to the way knack handles time.

I am trying to alert people who have said they would return at a certain time and haven’t. So the tasks runs when journey is still open and return time is before current time. People can select return time in 15 min intervals, so I have to run first task after 10 mins to give people some grace as clearly a few min delay is to be expected, but have to run tasks before the next quarter hour, in order to give those grace, so tasks currently run for example at xx:41 and xx:43.

I could I guess running a flag task first based with a different variable per each time it runs and a further task say 30 mins later looking for that single variable, but creating a set of 2 or more likely 3 sets of 96 tasks with no ability to sort or copy them is just too much pain, especially if I want to make changes.

Thanks for comments and for knack responding to support emails so thoroughly and promptly. It’s put it all in context.

One of the pitfalls of a no code solution!

I don’t think it’s a pitfall of nocode just a limitation on scheduling and running concurrent tasks on a shared server. :blush:

Have you looked at using Integromat or Zapier to run the schedule?

Normally Tasks are quite reliable. Use the all the time. Did have an issue once, where I had a lot of tasks … dozens … running, with some of the values they were dealing with were dependent the the outcome of previous tasks that were running. Tried to space the time out, but sometimes it just didn’t work. It was a while ago, but I recall Knack support indicating to me when what volume of tasks start being a potential issue.