Re: executing SELECT xmlelement(name foo); causes "server closed the connection unexpectedly" Error

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Peter Eisentraut <peter_e(at)gmx(dot)net>
Cc: Sushil <sdive(at)vertex(dot)co(dot)in>, pgsql-bugs(at)postgresql(dot)org, pgsqlrpms-hackers(at)pgfoundry(dot)org
Subject: Re: executing SELECT xmlelement(name foo); causes "server closed the connection unexpectedly" Error
Date: 2008-11-21 18:17:43
Message-ID: 10218.1227291463@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

Peter Eisentraut <peter_e(at)gmx(dot)net> writes:
> Sushil wrote:
>> *postgres: postgres testdb [local] SELECT: symbol lookup error:
>> postgres: postgres testdb [local] SELECT: undefined symbol:
>> xmlNewTextWriterMemory*

> Your problem appears to be here. Check you libxml installation. Maybe
> someone forgot to export this symbol.

I think it most likely is a corrupted download. Red Hat's normal
release process includes checks for ABI breakage such as disappearing
symbols, so it's hard to believe there's really a problem of that ilk.

The annoying thing about this from a Postgres standpoint is that the
missing symbol results in a runtime crash; a failure at server startup
would be a lot better IMHO. Ideally, shared libraries would be
processed with the equivalent of dlopen(RTLD_NOW). But as far as I
can find we can't easily force that on Linux --- the only available
way to determine it is to set a magic environment variable that the
linker consults.

I could fix this in the RPM distribution by putting export LD_BIND_NOW=1
into the postmaster start script, but I wonder whether there's a better
way.

regards, tom lane

In response to

Responses

Browse pgsql-bugs by date

  From Date Subject
Next Message Alexandra Nitzschke 2008-11-21 18:19:44 Re: could not read block 77 of relation 1663/16385/388818775
Previous Message Tom Lane 2008-11-21 17:35:31 Re: could not read block 77 of relation 1663/16385/388818775