On Mon, Jun 12, 2023 at 4:58 AM Jürgen Kuri
<[email protected]> wrote:
>
> Hello Gabriel, hello Nick & Mike,
>
> I'm also one the users which had problems with disconnects in an earlier 
> release (guacd 1.4.0). There were also core dumps with earlier releases 
> (1.2.0) in a different connection which I can't remember anymore (as far as I 
> recall for that I also created a ticket). At that time I opened a detailed 
> ticket and provided core dumps and backtraces.

Hopefully these issues have been taken care of in later releases - if
not, let us know on those mailing list threads and/or Jira issues and
we'll have another look.

>
> I don't know if got this already but for generating core dumps you have to 
> prepare your system. Normally, per default, writing core dumps to disk is 
> disabled (RLIMIT_CORE). Before you start a process, here guacd, you have to 
> enable it. So type ulimit -a (for all limits or -c, just seeing the core dump 
> limit) in your shell:

This depends to some extent on the distribution you're running. For
example, RHEL (/CentOS/Rocky) has the "abrt" daemon which, when
installed and active, can handle core dumps and collect those and
other information into a central location to aide in bug reporting.
Some distributions, like you point out, disable core dumps by default
and they have to be explicitly enabled.

>
> I'm really keen on hearing if the core dump problems are solved somewhen 
> since I experienced core dumps of guacd in different connections not only 
> what is discussed here. There were also some on weak connections with small 
> bandwidth in connection with audio (microphone).

I'd just suggest you track those on the existing conversations and/or
bug reports, and update appropriately. In the cases where we cannot
reproduce them ourselves we're relying on the information provided to
determine if they are in Guacamole or some other system component.

>
> Also intresting to hear to rebuild the FreeRDP library - did it help?
>

At this point, the only time we're providing this as an alternative is
with the issue with _aligned_recalloc(), and it isn't just rebuilding
the library, it's building a newer version that does not contain the
bug that causes the segfault.

-Nick

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to