> However, after that when I try to run a script to dump another database and
> restore it onto the beta, I get a "file not found" error, which is really
> odd both because it was working fine on the 2009-03-24 build and because the
> permissions have not changed. Aside from that, which is it's own problem,
> after that error the service is no longer running and has to be started
> again.
>
> This makes me think that something similar happens during the install, so
> that something fails with the plperl setup that causes the service to shut
> down (rather than it failing to start up in the first place.)
If I shut down the 8.4-beta1 service and start up the 2009-03-24
service on the same port and run the script the same way (it's a perl
script I'm accessing through a browser), it runs fine, so I can't see
how it would be an Apache or Perl permissions issue (i.e., what those
two are running as)--I just hit F5, it's launching the same pg_restore
binary from the same account on the same restore file. The one way is
fine, the other says, "No such file or directory" (sorry: not "file
not found," that was a bit imprecise of me earlier.)
I found I was looking in the wrong log for more detail, but I found
some:
pg_restore: WARNING: database "production" does not exist
pg_restore: [archiver (db)] Error while PROCESSING TOC:
pg_restore: [archiver (db)] Error from TOC entry 519; 2612 47275
PROCEDURAL LANGUAGE plperl mysuperuser
pg_restore: [archiver (db)] could not execute query: server closed the
connection unexpectedly
This probably means the server terminated abnormally
before or while processing the request.
Command was: CREATE PROCEDURAL LANGUAGE plperl;
After that it's an error with every statement because there's no
connection to the server.
Now the really weird thing is, "production" doesn't exist on my
2009-03-24 server either. That's the one the dump is *from*, but I'm
using "pg_restore -d dev", so it shouldn't matter that it doesn't
exist (and indeed it doesn't on the 2009-03-24 server, because that
works fine, note that I'm even using the 8.4-beta1 pg_restore binary
for both cases.) And for the record the script drops "dev" and
creates it from template0 right before trying to restore over it.
Is it possible it's looking for Perl in the wrong place? (Hence the
"No such file..." error that somehow makes it back to my Perl script?)
Kev