Yep. And even the memory leak should be fixed. It is not even static.

----- Original Message -----
From: "James Wang" <[EMAIL PROTECTED]>
To: "Aarno Syvanen" <[EMAIL PROTECTED]>; <[EMAIL PROTECTED]>
Sent: Friday, April 26, 2002 12:58 PM
Subject: Re: Fw: Kannel version 1.1.5-1 fault tolerance and memory leaks


>
> ----- Original Message -----
> From: "Aarno Syvanen" <[EMAIL PROTECTED]>
> > > >  ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> > > >  1). If I send an empty UDP packet to the WAP box, it does not
> > > >   serve any more.
> > > >   (I have to restart the service.)
> >
> > Can you report your settings with more details ?
>
> >followings are extracted from /var/log/kannel/wapbox.log
>
> >2002-04-19 12:41:51 [2] WARNING: WSP UNIT: Empty datagram.
> >2002-04-19 12:45:51 [2] WARNING: WSP UNIT: Empty datagram.
> >2002-04-19 12:51:51 [2] WARNING: WSP UNIT: Empty datagram.
> >2002-04-19 12:55:52 [2] WARNING: WSP UNIT: Empty datagram.
> >2002-04-19 12:56:20 [2] WARNING: WSP UNIT: Empty datagram.
>
> >It seems that connectionless WSP code logs empty datagram
> >properly but does not release memory in this case. Can
> >you fix this,
>
>  ^^^ means who?
>
>
> >having a stress test program for wap helps
> >in this regard.
>
> >fresh starting actually at:
> >               size        res
> >     Wapbox:  8048K        4160K
> >     Bearbox: 7680K        3760K
>
> > > >
> > > >     After 239,000 hits:
> > > >               size        res
> > > >     Wapbox:  8912K        5552K
> > > >     Bearbox: 10M          6040K
> >
> > Yes, bearerbox leak no hopefully fixed, thanks
>                       ^^
> >should be             now?
>
> >Yeah. I actually run stress tests one hour, 350 msg/s, with no non-static
> >leaks in bearerbox.
>
> So you have a new version now?
>
> >Wapbox leak is smallish, because empty
> >message is an exception.
>
> OK.  However, it does not serve anymore if
> being sent an empty UDP packet, which is
> fatal.
>
> Cheers
> James Wang
>
>
>
>
>
>


Reply via email to