Victor Duchovni <[EMAIL PROTECTED]> wrote: > This is not difficult to solve. Write a progressive parser that saves > state just for messages that are not yet done (still in the queue). For [...]
That's sort of what I described in my second paragraph. It's a workaround, but it's far from perfect, and some situations can cause it to get the wrong results unless you anticipate unusual situations. Each unusual situation has a possible fix, of course. This isn't impossible to work around. That doesn't stop me from wishing that it wasn't necessary. Wietse Venema <[EMAIL PROTECTED]> wrote: > There is no guarantee that mail goes from incoming->qmgr only once. > Whenever the qmgr is restarted, mail goes back from active queue > to incoming queue, so that it doesn't drown in a sea of deferred > mail. I know that, it's another example I could've added to the list of situations where remembering IDs seen in cleanup log messages can get tripped up. Even so the list would not be exhaustive. The point is, if qmgr logged where it got the message from, there'd be no need to think of all these contingencies. > If your problem is a large incoming queue, then I suggest that you > take the necessary step to find out why Postfix defers massive [...] I'm not describing a specific problem in this email thread, just a feature I occasionally wish for and why it could be useful. Yes, when there's a large incoming queue, that's a problem that needs to be solved, and I've run into this and solved it several times over the years. However, it's exactly when working on solving such a problem that log analysis & monitoring can be most useful, and the inability to easily distinguish between log entries from messages coming from incoming vs. deferred is a minor frustration. Obviously having qmgr add this bit of info to its log messages wouldn't actually solve any underlying problems. But it could help make it easier to solve them, or to monitor the progress of a possible solution, or at least get a handle on the extent of the problem when first noticing there's a problem. -- Cos
