3.0.6 strange bug causing infinite loop on "Ready to process requests" message

[Nick Rogers]

On Tue, Jan 6, 2015 at 3:28 PM, Arran Cudbard-Bell <
a.cudbardb at freeradius.org> wrote:

>
> > On 6 Jan 2015, at 17:02, Alan DeKok <aland at deployingradius.com> wrote:
> >
> > On Jan 6, 2015, at 4:48 PM, Nick Rogers <ncrogers at gmail.com> wrote:
> >> After sending the first test authentication request, generated by
> radtest client, radiusd emits a second "Ready to process requests" line and
> continues to repeat it over and over at will and without delay, until the
> log filesystem is full. Stopping it requires a kill -9. This seems to
> happen only when a request is received. The server does not send a response.
> >
> >  I don’t see that in my tests.  The debug output you posted also doesn’t
> show it receiving a packet.  So it’s hard to tell what’s going on.
> >
> >> The only thing unusual about my config is that I am using rlm_perl with
> threaded perl. However the server never seems to enter the rlm_perl module.
> >
> >  It should run the perl module if it receives a request.
> >
> >> I'm hoping someone can help identify if this is actually a bug or
> something dumb in my configuration, or a problem with my binary?
> >
> >  It looks like the binary is broken.  There’s no configuration option
> which says “misbehave horribly”.
>
> Run it under gdb or lldb, provide backtrace of all threads.
>
> Possibly related to the kevent changes as no Linux users have reported it.
> No guarantee kevent behaves
> identically on OSX and BSD for corner cases.
>

Yeah, that has been my gut feeling as well after seeing the kqueue addition
in the changelog. I've definitely experienced kqueue related issues with
other daemons under FreeBSD.

Here's the thread backtrace

Starting program: /usr/local/sbin/radiusd -fxx
(no debugging symbols found)...(no debugging symbols found)...(no debugging
symbols found)...(no debugging symbols found)...(no debugging symbols
found)...(no debugging symbols found)...(no debugging symbols found)...(no
debugging symbols found)...[New LWP 100270]
(no debugging symbols found)...(no debugging symbols found)...(no debugging
symbols found)...(no debugging symbols found)...(no debugging symbols
found)...(no debugging symbols found)...(no debugging symbols found)...(no
debugging symbols found)...(no debugging symbols found)...[New Thread
803406400 (LWP 100270/radiusd)]

Program received signal SIGINT, Interrupt.
[Switching to Thread 803406400 (LWP 100270/radiusd)]
0x0000000802164bda in write () from /lib/libc.so.7
* 2 Thread 803406400 (LWP 100270/radiusd)  0x0000000802164bda in write ()
from /lib/libc.so.7

Thread 2 (Thread 803406400 (LWP 100270/radiusd)):
#0  0x0000000802164bda in write () from /lib/libc.so.7
No symbol table info available.
#1  0x00000008019a3ef6 in write () from /lib/libthr.so.3
No symbol table info available.
#2  0x00000008008ab5f7 in vradlog () from
/usr/local/lib/freeradius-3.0.6/libfreeradius-server.so
No symbol table info available.
#3  0x00000008008ab83e in radlog () from
/usr/local/lib/freeradius-3.0.6/libfreeradius-server.so
No symbol table info available.
#4  0x00000000004428d5 in radius_event_init ()
No symbol table info available.
#5  0x0000000800b0cd54 in fr_event_loop () from
/usr/local/lib/freeradius-3.0.6/libfreeradius-radius.so
No symbol table info available.
#6  0x0000000000443571 in radius_event_process ()
No symbol table info available.
#7  0x0000000000433306 in main ()
No symbol table info available.
The program is running.  Exit anyway? (y or n) fbsd101-vm#

SIGINT was sent when it was stuck in the "Ready to process requests" loop.
Looks to me like it is at least event related, but I'm not sure?

This is with a very simple configuration, basically empty configuration
with no modules or server config. It doesn't even get as far as authorizing
the client.

I was working on trying to disable kqueue somehow and test. Any suggestions
on the best way of doing that? There does not appear to be a configure flag
for it, unless I am missing it.


> -Arran
>
> Arran Cudbard-Bell <a.cudbardb at freeradius.org>
> FreeRADIUS development team
>
> FD31 3077 42EC 7FCD 32FE 5EE2 56CF 27F9 30A8 CAA2
>
> -
> List info/subscribe/unsubscribe? See
> http://www.freeradius.org/list/users.html
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freeradius.org/pipermail/freeradius-users/attachments/20150106/8e49e469/attachment.html>