[Dnsmasq-discuss] segfaults in 2.39 after a while

[Joe Landman]

Hi Simon:

Simon Kelley wrote:

> I'm out beyond the wilder shores of Unix process-law here. Any idea 
> exactly what happens so that is can be guarded against? I'm unhappy with 
> the idea that it's possible to segfault the daemon any way.

Here is a WAG.  The parent spawns a child, which, while the parent is 
waiting on the child to complete, the child causes the parent to 
restart/die.  Which as the resources have not been freed (since the 
child process is still effectively locking the parent process, the dead 
one), the new process cannot effectively access the resources.

> HUP is fine as long as the configuration you are changing is in 
> /etc/ethers. The next release will go further and have the ability to 
> (re-)read a file which can contain complete dhcp-host type machine 
> information. That allows on-the-fly update of tags, client-id and lease 
> times, as well as MAC/IP address.

Cool.  I am dumping my cluster information into 
/etc/cluster/hosts.cluster.  All I want dnsmasq to do is to reread that 
file.  I think HUP does it, I was just being lazy with a restart (old 
code, replaced with a simpler bit of new code).

BTW:  the tftp server is quite nice.  Using it extensively, along with 
dhcp, and of course the dns functions.

> 
>>
>>
> 
> Cheers,
> 
> Simon.


-- 

Joseph Landman, Ph.D
Founder and CEO
Scalable Informatics LLC,
email: landman at scalableinformatics.com
web  : http://www.scalableinformatics.com
        http://jackrabbit.scalableinformatics.com
phone: +1 734 786 8423
fax  : +1 866 888 3112
cell : +1 734 612 4615