Re: BUG #16283: crash on create index segmentation fault - Mailing list pgsql-bugs

From Tomas Vondra
Subject Re: BUG #16283: crash on create index segmentation fault
Date
Msg-id 20200228210021.znum3xr3kprj3qec@development
Whole thread Raw
In response to Re: BUG #16283: crash on create index segmentation fault  (Przemysław Szustak <przemyslaw.szustak@gmail.com>)
Responses Re: BUG #16283: crash on create index segmentation fault  (Przemysław Szustak <przemyslaw.szustak@gmail.com>)
List pgsql-bugs
On Fri, Feb 28, 2020 at 08:58:19PM +0100, Przemysław Szustak wrote:
>I updated info on github logs because now system was generated new core
>dump.
>
>How can I get you more information? I don't now how to debug it.
>

Well, it's hard to say what exactly is happening, without getting a
better backtrace. Your system is apparently missing debug symbols, so
you need to do something like

   sudo aptitude install postgresql-10-dbg

or something like that. And the same for postgis, I assume. Then
generate the backtrace again (using the existing core file).

You might also attach gdb to the running backend, trigger the issue
and do more investigation there.

1) determine the PID of the backend handling your connection

   db=# select pg_backend_pid();

2) attach GDB to the process

   $ gdb -p $PID
   (gdb) c

3) Run the query using the same backend, once it triggers you'll be able
to do 'bt' in the gdb shell and inspect variables.

This needs the debug symbols too, though.


regards

-- 
Tomas Vondra                  http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services



pgsql-bugs by date:

Previous
From: Daniel Gustafsson
Date:
Subject: Re: BUG #16274: Repeated Libraries in Mac
Next
From: Dean Rasheed
Date:
Subject: Re: BUG #16281: LN() function inaccurate at 1000th fractional digit