Re: initdb failure with PostgreSQL 7.3.2 / Cygwin 1.3.22-1 / - Mailing list pgsql-cygwin
From | Andrew Bailey |
---|---|
Subject | Re: initdb failure with PostgreSQL 7.3.2 / Cygwin 1.3.22-1 / |
Date | |
Msg-id | 008301c3134b$8682f470$e800a8c0@PLANETNOMAD Whole thread Raw |
In response to | Re: initdb failure with PostgreSQL 7.3.2 / Cygwin 1.3.22-1 / (Frank Seesink <frank@mail.wvnet.edu>) |
List | pgsql-cygwin |
----- Original Message ----- From: "Frank Seesink" <frank@mail.wvnet.edu> To: <pgsql-cygwin@postgresql.org> Sent: Monday, May 05, 2003 8:36 PM Subject: Re: [CYGWIN] initdb failure with PostgreSQL 7.3.2 / Cygwin 1.3.22-1 / > Jason Tishler wrote: > > Frank, > > > > On Mon, May 05, 2003 at 08:13:01AM -0400, Jason Tishler wrote: > > > >>On Fri, May 02, 2003 at 06:11:06PM -0400, Frank Seesink wrote: > >> > >>>It's as if the copy of ipc-daemon running in the 'postgres' context > >>>is oblivious to the copy running as Local System account. > >> > >>Yes. I believe understanding the above will lead to the root cause of > >>this problem. > > > > > > I did some more Googling. Does the following apply to your setup? > > > > http://support.microsoft.com/default.aspx?scid=kb;en-us;264651 > > Jason, > > After my last post regarding the above link, I switched over to the > 'postgres' account and tried 'ipctest s' and it worked!!! GAAH! I > started feeling like Markko Paas (the thread that started on 16 Jan > 2003) which culminated with him posting on 22 Jan 2003 that > > ..."ipctest s" started working "out of blue" > > but I wanted to get to the bottom of this, once and for all. Then this > post of yours tickled my brain. > > Turns out you da man! I think you may have found the culprit! You > see, I had LOGGED OUT as 'Frank', then logged in as 'postgres'. To test > my theory, I logged out of 'postgres', logged back in as 'Frank', then > switched ([WindowKey]-L) out and logged in as 'postgres'. Sure enough, > 'ipctest s' failed! > > I have now tested further by logging back in as 'Frank' (needed admin > rights) and doing > > $ net stop ipc-daemon > $ rm /tmp/MultiFile* > > then rebooting (ipc-daemon is set as a service to start > 'Automatically'), logged in as 'postgres', and ran 'ipctest s'. It worked! > > Darn it all to heck, that stupid Windows Terminal Services issue is > STILL there!!! Now, I have no clue how you switch it from 'Remote > Administration' mode to 'Application Server' mode. But I believe the > very act of using the Fast User Switching (where you 'Switch Users' > without logging out) is what is causing this. > > I have tried this several times now, and it consistently points to the > fact that if you run ipctest when you have another account logged in > (and that's IT no less...no need to be running any Cywgin apps, the BASH > shell, whatnot), ipctest fails. I have no clue whether this would also > apply if someone used the VPN server feature in XP Pro and up, but note > that the 'Fast User Switch Compatibility' NT service (as well as others) > all rely on the 'Terminal Services' NT service, so what you may be > looking at here is a cascading effect: any action/service that relies > on 'Terminal Services' may trigger this gotcha. > > For a final test, I am going to blow away the entire Cygwin > distribution install, clean house on the Registry, etc., and start anew. > I will then install Cygwin as I always have done, and simply avoid > doing any user context switching. If I have any difficulties, I suspect > they may be with the default file permissions of the /tmp and /usr/bin > directories. I will simply do the steps as outlined in the PostgreSQL > README for starters, however, and once done, I'll let you know what > steps are required beyond the basic install (like possibly 'chmod 777 > /tmp' and 'chmod 755 /usr/bin /usr/bin/*'). > > This information may come in handy for others trying to install/run > PostgreSQL on Cywgin under Windows XP. I'd really appreciate clearer indication when to log in as myself (administrator), and when to log in as Postgres. > Again, thanks for the diligence in this, Jason. I'll try to > reciprocate the favor now in what little way I can (though may not post > 'til morning depending how long it takes me). I want to know this for > my own sake as much as for helping others. I don't want to see > PostgreSQL stop 'out of the blue' any more than the next guy. :-) > > > ---------------------------(end of broadcast)--------------------------- > TIP 4: Don't 'kill -9' the postmaster
pgsql-cygwin by date: