BUG #5682: Postgres Service crashes with exception 0xC0000135

On 09/29/2010 03:08 PM, Aswin J wrote:

The following bug has been logged online:

Bug reference: 5682
Logged by: Aswin J
Email address: jayaswin@gmail.com
PostgreSQL version: 8.3.7
Operating system: Windows

Windows ... what? XP? Vista? 7? 2003? 2008?
32- or 64-bit?

2010-09-27 10:33:30 IST LOG: server process (PID 22656) was terminated by
exception 0xC0000135
2010-09-27 10:33:30 IST HINT: See C include file "ntstatus.h" for a
description of the hexadecimal value.

"The application failed to initialize". Apparently usually due to
missing/bad runtime libraries, making this an odd thing to have arise
for a running Pg instance.

When does it crash? When forking a new backend (new connection) ? When
running a particular command? Randomly, when you're not doing anything?

2010-09-27 10:33:55 IST FATAL: the database system is in recovery mode
2010-09-27 10:33:55 IST LOG: all server processes terminated;
reinitializing
2010-09-27 10:33:56 IST FATAL: pre-existing shared memory block is still in
use
2010-09-27 10:33:56 IST HINT: Check if there are any old server processes
still running, and terminate them.

There were some fixes to shared memory on windows since 8.3. Does 9.0
work better on your machine?

--
Craig Ringer

Please reply with "reply all" so that others who're on the -bugs list
may see and respond to your posts.

I've CC'd the list. My reply follows below.

On 29/09/2010 4:32 PM, aswin jayaraman wrote:

I am running it on a Windows Vista 32 bit.The crash happens randomly.So
the connections we make to the database fail. This though is not seen on
all machines but on the ones it is seen it is quite consistent.Should I
try postgreSQL 9.0 for better results? Or is there a fix for the issue?

I haven't personally seen this particular issue reported, though there
have been plenty of problems with shared memory that seem to have been
fixed in 8.4 .

It would be a good idea to try 8.4 or 9.0 and see if you still have the
problem there. If you do, then we'll at least be working with a current
version when debugging/testing.

--
Craig Ringer

Tech-related writing at http://soapyfrogs.blogspot.com/

I ll try that out.Is there means to do an upgrade to 8.3 to 8.4/9.0 with the
data in place?

Thanks in Advance,
Aswin

On Wed, Sep 29, 2010 at 2:15 PM, Craig Ringer
<craig@postnewspapers.com.au>wrote:

On 29/09/2010 5:11 PM, aswin jayaraman wrote:

I ll try that out.Is there means to do an upgrade to 8.3 to 8.4/9.0 with
the data in place?

No, it requires a dump and reload. Sorry. You'll want to read the
release notes, as there have been changes between 8.3 and 9.0 that may
affect applications.

http://www.postgresql.org/docs/9/static/release-8-4.html
http://www.postgresql.org/docs/9/static/release-9-0.html

PostgreSQL has seen significant improvements in the Windows port, so
it's worth the update. I unfortunately cannot promise that it'll fix the
issue you're having, though.

--
Craig Ringer

Tech-related writing at http://soapyfrogs.blogspot.com/

Craig Ringer wrote:

On 29/09/2010 5:11 PM, aswin jayaraman wrote:

I ll try that out.Is there means to do an upgrade to 8.3 to 8.4/9.0 with
the data in place?

No, it requires a dump and reload. Sorry. You'll want to read the
release notes, as there have been changes between 8.3 and 9.0 that may
affect applications.

http://www.postgresql.org/docs/9/static/release-8-4.html
http://www.postgresql.org/docs/9/static/release-9-0.html

PostgreSQL has seen significant improvements in the Windows port, so
it's worth the update. I unfortunately cannot promise that it'll fix the
issue you're having, though.

pg_upgrade works for upgrades from 8.3 to 9.0.

--
Bruce Momjian <bruce@momjian.us> http://momjian.us
EnterpriseDB http://enterprisedb.com

+ It's impossible for everything to be true. +