Thread: 7.4 logging bug.

I just installed a 7.4 on windows/cygwin.  I restored a dump but
ran out of disk space during the creating of an index.  In psql I
saw the "ERROR: could not extend relation ....".

From that point on it seems to have stopped logging most things.

The ERROR and HINT are not in the log file, it stopped showing
the executed commands, and duration.  The only thing it
still shows is the "recycled transation log file" messages.

In psql I still see all the messages, like the notice that adding
a primary key will create an index.


Kurt

Kurt Roeckx <Q@ping.be> writes:
> I just installed a 7.4 on windows/cygwin.  I restored a dump but
> ran out of disk space during the creating of an index.  In psql I
> saw the "ERROR: could not extend relation ....".
> From that point on it seems to have stopped logging most things.

Do you suppose that the stdio support under cygwin somehow permanently
closes the stderr output descriptor after the first failure to write?
That would be unpleasant.  (Are you sure there *is* now space to write?)
        regards, tom lane

On Thu, Nov 20, 2003 at 04:08:28PM -0500, Tom Lane wrote:
> Kurt Roeckx <Q@ping.be> writes:
> > I just installed a 7.4 on windows/cygwin.  I restored a dump but
> > ran out of disk space during the creating of an index.  In psql I
> > saw the "ERROR: could not extend relation ....".
> > From that point on it seems to have stopped logging most things.
> 
> Do you suppose that the stdio support under cygwin somehow permanently
> closes the stderr output descriptor after the first failure to write?
> That would be unpleasant.  (Are you sure there *is* now space to write?)

There is now space to write yes.  I just started to delete some
things at the moment it ran out of disk space.

It's still logging the "recycled transation log file".  Is that
send to stdout instead of stderr maybe?


Kurt

Kurt Roeckx <Q@ping.be> writes:
> It's still logging the "recycled transation log file".  Is that
> send to stdout instead of stderr maybe?

No, it all goes to stderr.  But that output comes from a different
subprocess.  Not sure why that subprocess would still have working
stderr if others don't ... any ideas anyone?

I'd suggest you restart the postmaster and see if things get better,
except I'm afraid of destroying the evidence.

Does cygwin have any equivalent of strace/ktrace?  It'd be useful
to see whether the write() calls are still being issued or not.
        regards, tom lane

On Fri, Nov 21, 2003 at 02:49:28AM -0500, Tom Lane wrote:
> Kurt Roeckx <Q@ping.be> writes:
> > It's still logging the "recycled transation log file".  Is that
> > send to stdout instead of stderr maybe?
> 
> No, it all goes to stderr.  But that output comes from a different
> subprocess.  Not sure why that subprocess would still have working
> stderr if others don't ... any ideas anyone?

It seems quiting and starting psql "fixed" it.  So I guess it's
the backend that had the problem.  So it was a different
subprocess.

> Does cygwin have any equivalent of strace/ktrace?  It'd be useful
> to see whether the write() calls are still being issued or not.

It has a strace, but when I use it I get a nice exception.  I
never really got either strace or gdb to work properly under
cygwin when attachting to an existing process.


Kurt