Re: FATAL: invalid value for option 'LC_MESSAGES': 'en_US' - Mailing list pgsql-cygwin

    Never mind.  In short, yes it was a simple problem.

    Turns out that if you have an environment variable LANG set (e.g.,
LANG=en_US), this wreaks havoc when you run initdb.  I'm afraid I don't
know why (though I'll be digging into that).  But in short, my 2nd PC
had that variable set (due to an IBM PL/1 compiler installation on that
machine...looooong story) whereas the test PC did not.

    Standard Windows XP installs don't have this variable set, but just in
case you have a similar problem, it's an easy fix.

    Just before executing the 'initdb' command, do the following:

    $ set | more

and look to see if you have a LANG environment variable set.  Cygwin
will not have this by default, so it will have sucked it in from the
Windows environment table if it's there.  If so, simply type

    $ unset LANG

to unset it within that BASH session, then do your 'initdb' as usual.
Don't worry.  This has no long-lasting effects.  The only way to remove
that environment variable for good, you need to remove it from Windows
by RIGHT-clicking on 'My Computer' & clicking 'Properties' from the
popup, then navigating to the 'Advanced' tab,
[Environment Variables], highlighting the LANG environment variable
(most likely in System Variables) and clicking [Delete], then [Ok]ing
your way out.

    Anyway, problem solved.  Once initdb is run, this environment variable
doesn't seem to get in the way (but don't quote me on that).


Frank Seesink wrote:
> Alright, new issue...hope it's minor.
>
> Installed Cygwin and PostgreSQL clean following my own instructions
> posted to this list, but when I get to the step to fire up the
> PostgreSQL engine itself via
>
>     $ net start postmaster
>
> I get the gloriously useless
>     ____________________________________________________________
>     The postmaster service is starting.
>     The postmaster service could not be started.
>
>     The service did not report an error.
>
>     More help is available by typing NET HELPMSG 3534.
>     ____________________________________________________________
>
> And for anyone who has ever typed in "NET HELPMSG 3534", you know that
> you end up with "The service did not report an error".
>
>     Note this install is identical in every respect but one with the
> install I did on my test box, which worked just great.  The one
> difference is that I switched in the Cygwin1.dll nightly snapshot right
> in the beginning of this install (so it was in place when I did
> 'initdb'), whereas in the initial install I only dropped that in AFTER
> having trouble getting psql to connect via sockets.  However, I have
> ruled that issue out by restoring the distribution cygwin1.dll,
> resetting CygIPC, deleting the /usr/share/postgresql/data directory, and
> then redoing the 'initdb' command to initialize the database (while
> using the distribution version of cygwin1.dll), and in either case, I
> end up with postmaster failing to start.
>
>     As for the useless NET HELPMSG info, being the handy-dandy Windows
> guy that I am (translation:  I have to deal with this crap regularly), I
> checked the Application Event Log, where the error message gave me a
> little more insight in that PostgreSQL claims it did not have privilege
> to execute...though it was vague on what exactly.
>
>     To be safe, I re-typed the password for the 'postgres' account in
> the 'postmaster' NT service, but no luck.
>
>     I then checked /var/log/postmaster.log, which contained
>     ____________________________________________________________
>     FATAL:  invalid value for option 'LC_MESSAGES': 'en_US'
>     ____________________________________________________________
>
> More research led me to the file
>     /usr/share/postgresql/data/postgresql.conf
>
> at the end of which is just such a setting for LC_MESSAGES.  But that's
> as far as I get.
>
>     I'm still scratching my head as to why one box works just fine and
> the other does not, and why this particular message.  Thoughts?  Ideas?
>
> P.S.    Though once again, it should not play a part, I thought I better
>         mention that this new PC is running 2 CPUs, whereas my test box
>         only had 1.  Could this be an issue?  Is PostgreSQL sensitive
>     to dual CPU boxes?
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 6: Have you searched our list archives?
>
> http://archives.postgresql.org
>