Horton

Re: Server crash due to SIGBUS(Bus Error) when trying to access the memory created using dsm_create().

Thomas Munro <thomas.munro@enterprisedb.com>

View thread View on postgr.es
From: Thomas Munro <thomas.munro@enterprisedb.com>
To: Andres Freund <andres@anarazel.de>
Cc: Robert Haas <robertmhaas@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, amul sul <sul_amul@yahoo.co.in>, PostgreSQL-development <pgsql-hackers@postgresql.org>
Date: 2017-08-16T23:39:14Z
Lists: pgsql-hackers 
On Thu, Jun 29, 2017 at 12:24 PM, Thomas Munro
<thomas.munro@enterprisedb.com> wrote:
> fallocate-v5.patch

Added to commitfest so we don't lose track of this.

I'm mainly concerned about the fact that we have a way for PostgreSQL
to die that looks exactly like a bug, when really it's masking an
out-of-memory condition that a DBA or sysadmin would normally be able
to diagnose by the appearance of the OOM reaper at the window holding
a scythe.  I suppose the SIGBUS case is much more likely to happen
when we start actively using large amounts of dynamic shared memory
(parallel hash), but I suppose it could happen now if your system is
already overcommitted and a small DSM segment happens to push you over
the edge.

-- 
Thomas Munro
http://www.enterprisedb.com

Commits

  1. Fix failure-to-read-man-page in commit 899bd785c.

  2. Avoid SIGBUS on Linux when a DSM memory request overruns tmpfs.