Can I rely on trigger events always containing the state of the document at the exact time of the update?

MongoDB Atlas
1

This is a somewhat a follow-up question to Include metadata in Atlas trigger event data - for which I did not get any answers yet most likely because there is no answer.

So, a theoretical option would be to store the kind of “metadata” into the document and hence “know” who caused an event and which events belong together.

Now, the question is: Will this be reliable?

Like if user A updates document 1, setting changeId to abc and a microsecond later user A updates document 1, setting changeId to def, can I 100% of the cases rely on getting two events like

  • document 1 changed, current state of document: {"changeId": "abc"}
  • document 1 changed, current state of document: {"changeId": "def"}
    ?
    (In any random order.)

I don’t know next to nothing about MongoDB’s internals, but my assumption is that triggers work based on log watchers. And then the question would be: Do the logs contain the current state of the document? Or does MongoDB refetch the document from the DB? In the latter case, the state of the document could be outdated the moment it gets refetched and I would get twice

  • document 1 changed, current state of document: {"changeId": "def"}

and never

  • document 1 changed, current state of document: {"changeId": "abc"}