--- On Mon, 10/27/08, Mihai Zsigmond <[EMAIL PROTECTED]> wrote:

> From: Mihai Zsigmond <[EMAIL PROTECTED]>
> Subject: RE: Kannel startup script does not stop bearerbox completely
> To: [EMAIL PROTECTED]
> Date: Monday, October 27, 2008, 7:55 PM
> Hi guys,
> 
> I'm adding my 5 cents' worth at this discussion. 
> 
> The bearerbox waits for all SMSC connections to be properly
> closed, before it can close itself. This can take some time,
> mostly when connections are UCP/EMI type. You can skip this
> wait issuing a second kill command, which closes immediately
> the bearerbox and puts "Can not die at its own
> will" message in the log.
> 
> However, there is a more elegant approach to log rotation,
> which does NOT stop the bearerbox. 
> Instead, my log script renames the log file to some other
> name and then copies a blank file with the name of the log.
> This method has been working for the past four years on
> different Kannel versions on Fedora and CentOS systems.
> 
> Yet, I have never tested this method in high volume traffic
> environment, we are limited by our SMS providers to 1
> SMS/sec rate.
> 
> Hope it helps.
> 
> Best regards,
> Mihai Zsigmond
> 
> --- On Mon, 10/27/08, Mathieu Bruneau
> <[EMAIL PROTECTED]> wrote:
> 
> > From: Mathieu Bruneau
> <[EMAIL PROTECTED]>
> > Subject: RE: Kannel startup script does not stop
> bearerbox completely
> > To: "Jason Mule"
> <[EMAIL PROTECTED]>, [email protected]
> > Date: Monday, October 27, 2008, 6:36 PM
> > We have seen a similar thing, however ours _usually_
> > shutdown within
> > 30s... The SMS server is still on Debian Sarge
> however.
> > 
> > I usually do an strace on the process and see it's
> > actually does a futex
> > (Don't have the exact line atm) operation. I
> always
> > tought it was
> > actually waiting for all connection to be cleanly
> closed
> > (?). I'm
> > usually able to use the "status" command
> till
> > it's almost closed and see
> > which connections are still active.
> > 
> > However for log rotation, in sarge at least it's
> doing
> > an HUP and we
> > never had issue I can recall...
> > 
> > /var/log/kannel/*.log {
> >         daily
> >         missingok
> >         rotate 100
> >         olddir /var/log/kannel/backup
> >         compress
> >         create 640 kannel adm
> >         sharedscripts
> >         postrotate
> >                 killall -HUP bearerbox smsbox wapbox
> >
> > /dev/null 2>
> > /dev/null || true
> >         endscript
> > }
> > 
> > Regards,
> > Mathieu Bruneau
> > 
> > -----Original Message-----
> > From: Jason Mule [mailto:[EMAIL PROTECTED] 
> > Sent: Monday, October 27, 2008 11:02 AM
> > To: [email protected]
> > Subject: Kannel startup script does not stop bearerbox
> > completely
> > 
> > Hi,
> > Can anyone else on this list confirm this? I have
> noticed
> > that the
> > Kannel init script (Debian Etch, Kannel 1.4.1) does
> not
> > stop bearerbox
> > completely when called. After calling
> > 'start-stop-daemon --stop --retry
> > 5 --pidfile $PIDFILES/kannel_bearerbox.pid --exec
> > $BOXPATH/run_kannel_box' , a bearerbox process
> remains
> > which must be
> > killed manually or by adding a line similar to the one
> > below to the init
> > script:
> > 
> > # Wait for bearerbox to finish
> > start-stop-daemon --stop --quiet --oknodo
> > --retry=0/30/KILL/5 --exec
> > $BOXPATH/bearerbox
> > 
> > --
> > Kind regards
> > Jason Mule


      

Reply via email to