qpid-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Bill Freeman <ke1g...@gmail.com>
Subject Re: Maybe bug, maybe novice mistake, or maybe my python qpid library is too old.
Date Wed, 10 Apr 2013 20:42:55 GMT
On Wed, Apr 10, 2013 at 12:42 PM, Ken Giusti <kgiusti@redhat.com> wrote:

>
> ----- Original Message -----
> > From: "Bill Freeman" <ke1g.nh@gmail.com>
> > To: "users" <users@qpid.apache.org>
> > Sent: Wednesday, April 10, 2013 11:23:04 AM
> > Subject: Re: Maybe bug, maybe novice mistake, or maybe my python qpid
> library is too old.
> >
> > Ken,
> >
>
> <snip>
>
> >
> > As long as I have your attention, does ;{mode:browse} affect the need to
> > acknowledge messages?
> >
> > Thanks, Bill
> >
>
>
> Hi Bill -
>
> Since you are not transferring ownership of the message, ack'ing each
> message isn't strictly necessary.
>
> However, I seem to recall that issuing an occasional ack could affect
> other aspects of the consumer, like replenishment of prefetch credit, or
> something else that may improve overall performance.
>
> But I'm not the best person to ask - anyone else care to weigh in?
>
> thanks,
>

I thought I saw somewhere in the code that there is a list of unacked
messages in the python client library, and things go there without any kind
of check, that I could see, as to whether they were a browse.  So it might
be that the broker doesn't need it, but that it consumes memory in the
client.  Of course I can't remember whether I saw that in the low level or
the messaging layer code.

If the broker doesn't need it, it's a pity to waste the bandwidth sending
the ack, assuming that client doesn't figure out that the ack isn't needed,
and quash it.

Bill

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message