[altq 716] Re: altq 3 and openbsd 2.8

[Norman Widders <torqumada@paladincorp.com.au>]

On Thu, 1 Feb 2001, Stefan Laudat wrote:

hmm, the box still locks up with a different config, i tried
putting maxburst 1 at the end of each line too, to no avail, any other
suggestions are welcome, it ran for 15mins, but then i tried sending a 1mb
email from another box, routing through the openbsd box, it immediately
froze, tips/advice/ideas, have i overlooked anything?

/Norman

interface ppp0 bandwidth 33K cbq
class cbq ppp0 root NULL pbandwidth 100 maxburst 1
#
# meta classes
#

class cbq ppp0 ctl_class root pbandwidth 4 control

# CBQ/TYPE, Iface, Name, Parent/Null
class cbq ppp0 fast_class root borrow pbandwidth 50 
        filter ppp0 fast_class 0 0 0 53 17      # dns
        filter ppp0 fast_class 0 0 0 53 6       # dns
        filter ppp0 fast_class 0 0 0 6667 6    # irc   
        filter ppp0 fast_class 0 6667 0 0 6    # irc
        filter ppp0 fast_class 0 0 0 22 6      # ssh
        filter ppp0 fast_class 0 22 0 0 6      # ssh

class cbq ppp0 slow_class root borrow pbandwidth 25 default
        filter ppp0 slow_class 0 0 0 25 6    # smtp
        filter ppp0 slow_class 0 25 0 0 6    # smtp
        filter ppp0 slow_class 0 0 0 80 6     # http    
        filter ppp0 slow_class 0 80 0 0 6     # http
        filter ppp0 slow_class 0 0 0 21 6      # ftp
        filter ppp0 slow_class 0 21 0 0 6      # ftp
        filter ppp0 slow_class 0 0 0 20 6      # ftp-data 
        filter ppp0 slow_class 0 20 0 0 6      # ftp-data
        filter ppp0 slow_class 0 0 0 110 6    # pop3
        filter ppp0 slow_class 0 110 0 0 6    # pop3   



> I've been through this quite some time ago...
> You can reproduce this nicely running altqd BEFORE bringing up device ppp0.
> As the server on that date had no videocard I didn't have the opportunity
> to switch on debugger and make a stack dump (sorry Kenjiro :(( ).
> You can use altq+ppp BUT you have to reboot the machine and start altqd AFTER
> the ppp device is up. Or maybe using ppp as kernel module and trying to
> unload/load it after a ppp failure.
> 
> 
> > wonder what i've done wrong, ever seen a lockup using altqd + openbsd 2.8,
> > happened twice, so i disabled altqd and went back to userland-ppp until i
> > find out why. it runs fine for a few minutes as altqstats shows
> > me.. previous uptimes were 100days before altq
> > 
> > not sure i am still experimenting and trying to get to grips with it
> > any tips/tricks would be welcome, as would an ICMP rule that lets me vary
> > it and see if having it is the problem? (lack of)
> > 
> > lo1: flags=8008<LOOPBACK,MULTICAST> mtu 32972
> > ne3: flags=8863<UP,BROADCAST,NOTRAILERS,RUNNING,SIMPLEX,MULTICAST> mtu
> > 1500
> >         media: Ethernet autoselect (10baseT)
> > de0: flags=8863<UP,BROADCAST,NOTRAILERS,RUNNING,SIMPLEX,MULTICAST> mtu
> > 1500
> >         media: Ethernet autoselect (10baseT)
> >         status: active
> >         inet 203.164.10.230 netmask 0xffffff80 broadcast 203.164.10.255
> > ppp0: flags=8010<POINTOPOINT,MULTICAST> mtu 1500
> >   ^^^^
> > swap ppp0 and tun0 here as i had kernel-pppd with altqd when it was
> > working/locked up.. i think i can duplicate it but prefer not to as having
> > a link go down hurts/ouch.
> > 
> > tun0: flags=11<UP,POINTOPOINT> mtu 1500
> >         inet 203.101.1.142 
> > 
> -- 
> Stefan Laudat
> -------------
> The only thing I feel when I kill is recoil.
> 

Paladin Corporation Pty Ltd.  Ph:+612 9835-4782     Fax:+612 9864-0487
Software Engineering:    c/c++/perl/sql/eiffel/pascal/haskell/php/java
Powered by FreeBSD/SMP                  http://www.paladincorp.com.au/
The  lyf  so  short,  the  craft  so  long  to  lerne     -    Chaucer
Unix, VMS, MVS since the 80's,   FreeBSD from 2.1.6,   Linux since '94