Re: radiusd doesn't dump a core after SIGSEGV?

[Coy Hile (BLOOMBERG/ 919 3RD A)]

From: aland at deployingradius.com At: 10/22/24 13:55:50 UTC-4:00To:  Coy Hile (BLOOMBERG/ 919 3RD A ) ,  freeradius-users at lists.freeradius.org
Subject: Re: radiusd doesn't dump a core after SIGSEGV?

On Oct 22, 2024, at 1:51 PM, Coy Hile (BLOOMBERG/ 919 3RD A) 
<chile1 at bloomberg.net> wrote:
> 
>> We've found that radiusd has died a couple times recently (using RedHat's 
packages on RHEL 8.8) without leaving a core file or even a stack trace that we 
can use to investigate the issue. Is that expected? I see the panic_action in 
`radiusd.conf`, but I don't want to run `gdb` on the core; I just want a core 
left behind so I can investigate the cause of the crash.
>> 
>> Any idea what I'm missing?

>  See radiusd.conf.  "allow_core_dumps"

 > The problem is that the OS often has additional settings which limit core 
dumps for background daemons.


Thanks, allow_core_dumps is what I didn't recall. Thanks, Alan.


>  And which version are you running?


>  But also... if you're using RedHat and paying for support, they should really 
be helping you.  We often find that RH distributions ship very old / out of 
date packages.


Yeah, they're still calling it 3.0.20-<something>, but they've cherry-picked/backported parts of 3.0.27 (*sigh*). I really want to get the NetworkRADIUS packages in here (especially since RedHat have deprecated functionality upon which we depend), but that's not immediate thing I can do.

 > Alan DeKok.