From: | Robert Haas <robertmhaas(at)gmail(dot)com> |
---|---|
To: | gradgis(dot)specker(at)bluewin(dot)ch |
Cc: | pgsql-bugs(at)postgresql(dot)org |
Subject: | Re: Failed to run initdb - not resolved bug 5130 |
Date: | 2010-01-15 16:51:34 |
Message-ID: | 603c8f071001150851q4307f175w3cc1902015e4fc48@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-bugs |
On Tue, Jan 12, 2010 at 8:04 AM, Thomas Specker
<gradgis(dot)specker(at)bluewin(dot)ch> wrote:
> Hi all
> after a long time of trying to install PostgreSQL with ...
>
> Computer:
>
> - Windows2000 (Jurassic Parc, I know ...)
>
> - PostGreSQL 8.3.9 and msi-file
> or PostGreSQL 8.4.2 and one click installer
>
> - Data in
> /8.4/data (default path)
> or D:\PostgresDaten (with 1 level)
> or D:\Postgres\PostgreSQLDaten (parent dir createt by me)
>
> Testing race:
> The result of all rounds is always the same:
>
> - Installation:
> runns through, but is not complete with error, when trying to initate the
> db: no postgresql.conf-file.
> The programdirectorys under c:\Programme\PostgreSQL\8.4 (or 8.3) are created
> and seem to be complete.
> But there are no datadirectorys or datafiles createt.
>
> - Running directly initdb.exe:
> stops with error "access denied" when trying to create something (for the
> template database) in \PostgreSQLDaten\base\1\.
> (\bin\initdb.exe -U postgres -D D:\Postgres\PostgreSQLDaten -W -n
> The parameter -n prevents from deleting of the createt files/dirs in case of
> errors.)
>
> Some datadirectorys, a postgresql.conf-file, a PG_VERSION-file, others
> *.conf-files are createt but the
> \base and \base\1\ are empty.
>
> - running pg_ctl.exe to start the server:
> complains about missing file pg_control and stopps.
> (\bin\pg_ctl.exe start -w -t 10 -D D:\Postgres\PostgreSQLDaten -l
> D:\Postgres\PostgreSQLKonfigInfos\logfile02.txt.
> Error in logfile is: D:/Postgres/PostgreSQLDaten/global/pg_control ... No
> such file or directory.
> This pg_ctl.exe wants the path- and filenames written exactly, with upper
> and lower cases. Only pg_ctl.exe ?)
>
> Somme more results:
>
> - In windows, the createt datadirectorys belong to the windows-user
> "thomas", with permissions for "administrator" and "postgres",
> but not allways (I missed to register this in every round of the race ...).
> - When running the initdb.exe without parameter -n, it allways could remove
> all artefacts.
> - on http://forums.enterprisedb.com/posts/list/1802.page can be seen, that
> running initdb.exe sometimes works.
> - The Windows-user postgres can create files and dirs in [\data]\base\1\.
>
> Diagnosis:
>
> The initdb is not allowed write to [\data]\base\1\ ("access denied"), but
> could create it.
> It is most likely, that the bug is inside initdb.exe.
> You can guess, that there is a conflict between windows-permissions and
> db-permissions.
> Windows-version, PostgreSQL-version, paths, locals, ... all seem not to
> matter.
>
> ... I now must hope.
>
> Thomas Specker
>
> If somebody has an installation with a SMALL, empty, and working
> data-section including the template-db, this could be tested (if just
> copying is possible).
This email is interesting in that it has a Reply-To header pointing to
"robertmhaas(at)gmail(dot)com". I haven't had any replies, but I wonder why
the poster thought that they should come to me.
On the actual content of the email, there's not really enough here for
anyone to begin speculating what the problem might be. Including the
actual error logs instead of summaries of what they contain would
probably be a good idea. The contention that "Windows-version,
PostgreSQL-version, path, locals, ... all seem not to matter" does not
seem likely to be true, because this does work for other people. But
without knowing specifically what was tried, it's hard for anyone to
help.
...Robert
From | Date | Subject | |
---|---|---|---|
Next Message | Jodi Escalante | 2010-01-15 17:23:02 | BUG #5281: Timestamp fields not inserting from 8.3 to 8.4 |
Previous Message | Kevin Grittner | 2010-01-15 16:45:09 | Re: BUG #5278: Postgres hangs / crashes every day |