Re: Slow downstream site wedged my Postfix...

Subject: Re: Slow downstream site wedged my Postfix...
From: Andrew McNamara (andrewmconnect.com.au)
Date: Wed Jan 26 2000 - 06:16:13 CST

• Next message: furio ercolessi: "Re: qmgr logging breaks"
• Previous message: Leif Nixon: "Re: qmgr logging breaks"
• In reply to: Wietse Venema: "Re: Slow downstream site wedged my Postfix..."
• Next in thread: Wietse Venema: "Re: Slow downstream site wedged my Postfix..."
• Next in thread: Matthias Andree: "Re: Slow downstream site wedged my Postfix..."
• Reply: Andrew McNamara: "Re: Slow downstream site wedged my Postfix..."
• Reply: Wietse Venema: "Re: Slow downstream site wedged my Postfix..."
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

>Perhaps it makes sense to combine delays (configurable) with
>something on the outbound side. Detecting that one destination
>hogs the queue is easy enough, adding a configurable switch to
>defer excess mail for a "hog" destination is not hard either.

I agree - having both would be valuable. The incident I referred to
yesterday was well within our machine's ability to handle - except they
wedged the active queue (I'm happy enough to queue 10000 jobs for one
customer for a finite period provided it doesn't effect other
customers).

Once I forcibly moved their jobs to the deferred queue, mail for other
sites flowed smoothly, although qmgr moved them back into active slowly
- I'm concerned that the proposal I suggested yesterday would just
delay the inevitable - qmgr would have to apply the same limits when
moving jobs from deferred as it applied moving them from incoming.

My thinking at the moment is that when processing jobs for recipients
sites that are over the limit, qmgr would shift jobs in incoming
straight to deferred, and ignore jobs in deferred for that site.

Wietse - can you comment on what I proposed (is it fatally flawed, any
gotchas in implementation)? I think I know how to do it, although my
understanding of postfix is rather blinkered.

An immediate problem that occurs to me is jobs with multiple
recipients. Hmmm - makes the problem resemble the ETRN problem.

 ---
Andrew McNamara (System Architect)

connect.com.au Pty Ltd
Lvl 3, 213 Miller St, North Sydney, NSW 2060, Australia
Phone: +61 2 9409 2117, Fax: +61 2 9409 2111

• Next message: furio ercolessi: "Re: qmgr logging breaks"
• Previous message: Leif Nixon: "Re: qmgr logging breaks"
• In reply to: Wietse Venema: "Re: Slow downstream site wedged my Postfix..."
• Next in thread: Wietse Venema: "Re: Slow downstream site wedged my Postfix..."
• Next in thread: Matthias Andree: "Re: Slow downstream site wedged my Postfix..."
• Reply: Andrew McNamara: "Re: Slow downstream site wedged my Postfix..."
• Reply: Wietse Venema: "Re: Slow downstream site wedged my Postfix..."
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

This archive was generated by hypermail 2b27 : Wed Jan 26 2000 - 06:21:13 CST