I have a booking API which creates sessions in my DB with one of the columns being scheduled check-in time. I need a functionality which auto checks in the user based on the scheduled check-in time. How do I go about it? The only solution I can think of is running a time-based job scheduler every minute which checks in all the user scheduled for that minute. Is there any better way to achieve this?

If it is likely that there is some work every minute, a scheduled job is not that bad. And even if it is not that likely using e constant scheduling interval might be a adequate choice because it is very reliable and usually has a better worst case behavior when many items are ready (because of the roughly constant load).

But the concept "every minute" has some lack. First of all if the job does not run once events in this minute will be lost forever. And if the job takes more than one minute to complete similar problems may arise.

So a better concept is, to execute all items not yet processed with a check-in time up to now, once the job is running. This will safely fall on its feet in case of some incidents.

However, if you want event based execution, i.e. no running job until an item is ready to be processed, you will necessarily need some kind of daemon running in background which schedules the tasks when ready. Some databases have built-in schedulers that can do this job. The Unix command at may do the job as well. (Be aware of time zone issues.)

So assuming you have some daemon that can execute a job at a given time then you need to feed this daemon. Every time a database update causes the next check-in time to change you need to adjust the time of the next job to start. A database trigger might be used for this purpose. But you need to be aware of race conditions here because concurrent updates may trigger in parallel. You also need to take care about lost starts because of some unexpected system failure or for whatever reason. I.e. some kind of monitoring is required.

I would not recommend to start a new job for every item to avoid the above complexity because this might cause excessive resource usage. And furthermore it requires even more monitoring.

Conclusion

Unless there are good reasons not to do so a cyclic scheduled job that checks for work is a simple and reliable solution.

The more complex way to schedule only when there is something to process is recommended if power consumption is an issue and/or the scheduling interval needs to be very small to hit the scheduling time very exactly. E.g. the OS kernel uses this technique to avoid unnecessary wake-ups of the CPU, but here we are talking about an accuracy within the milliseconds range.