Thread: Problems With -9.0.1

Problems With -9.0.1

From
Rich Shepard
Date:
   I built and installed -9.0.1 on my Slackware-13.1 server and it ran for a
while. Now, without warning, one of my applications dependent upon postgres
won't run. While trying to restart postgres I learned the .pid file did not
exist so I removed /tmp/.s.PGSQL* and tried to restart postgres.

   Here's what I see:

postgres@salmo:$ postgres -D /usr/local/pgsql/data &
[1] 16044
postgres@salmo:/var/log/httpd$ LOG:  could not open directory
"/share/timezone": No such file or directory
LOG:  could not open directory "/share/timezone": No such file or directory
LOG:  could not open directory "/share/timezone": No such file or directory
LOG:  could not open directory "/share/timezone": No such file or directory
LOG:  could not open directory "/share/timezone": No such file or directory
WARNING:  could not open directory "/share/timezonesets": No such file or
directory
HINT:  This may indicate an incomplete PostgreSQL installation, or that the
file "/bin/postgres" has been moved away from its proper location.
FATAL:  invalid value for parameter "timezone_abbreviations": "Default"

   The files and directories are in /usr/local/pgsql/share:

postgres@salmo:$ ls /usr/local/pgsql/share/
conversion_create.sql   postgres.bki            sql_features.txt
doc/                    postgres.description    system_views.sql
information_schema.sql  postgres.shdescription  timezone/
man/                    postgresql.conf.sample  timezonesets/
pg_hba.conf.sample      psqlrc.sample           tsearch_data/
pg_ident.conf.sample    recovery.conf.sample    unknown.pltcl
pg_service.conf.sample  snowball_create.sql

   What do I do to overcome this impasse and get postgres running again?

Rich

Re: Problems With -9.0.1

From
Adrian Klaver
Date:
On Monday 20 December 2010 5:09:24 pm Rich Shepard wrote:
>    I built and installed -9.0.1 on my Slackware-13.1 server and it ran for
> a while. Now, without warning, one of my applications dependent upon
> postgres won't run. While trying to restart postgres I learned the .pid
> file did not exist so I removed /tmp/.s.PGSQL* and tried to restart
> postgres.
>
>    Here's what I see:
>
> postgres@salmo:$ postgres -D /usr/local/pgsql/data &
> [1] 16044
> postgres@salmo:/var/log/httpd$ LOG:  could not open directory
> "/share/timezone": No such file or directory
> LOG:  could not open directory "/share/timezone": No such file or directory
> LOG:  could not open directory "/share/timezone": No such file or directory
> LOG:  could not open directory "/share/timezone": No such file or directory
> LOG:  could not open directory "/share/timezone": No such file or directory
> WARNING:  could not open directory "/share/timezonesets": No such file or
> directory
> HINT:  This may indicate an incomplete PostgreSQL installation, or that the
> file "/bin/postgres" has been moved away from its proper location.
> FATAL:  invalid value for parameter "timezone_abbreviations": "Default"
>
>    The files and directories are in /usr/local/pgsql/share:
>
> postgres@salmo:$ ls /usr/local/pgsql/share/
> conversion_create.sql   postgres.bki            sql_features.txt
> doc/                    postgres.description    system_views.sql
> information_schema.sql  postgres.shdescription  timezone/
> man/                    postgresql.conf.sample  timezonesets/
> pg_hba.conf.sample      psqlrc.sample           tsearch_data/
> pg_ident.conf.sample    recovery.conf.sample    unknown.pltcl
> pg_service.conf.sample  snowball_create.sql
>
>    What do I do to overcome this impasse and get postgres running again?
>
> Rich

As I remember it there was more than one version of Postgres on this machine.
Are you sure you are using the correct postgres binary? While I am it is there
a reason you are not using the system start scripts or pg_ctl:)?


--
Adrian Klaver
adrian.klaver@gmail.com

Re: Problems With -9.0.1

From
John R Pierce
Date:
On 12/20/10 5:09 PM, Rich Shepard wrote:
>   I built and installed -9.0.1 on my Slackware-13.1 server and it ran
> for a
> while. Now, without warning, one of my applications dependent upon
> postgres
> won't run. While trying to restart postgres I learned the .pid file
> did not
> exist so I removed /tmp/.s.PGSQL* and tried to restart postgres.
>
>   Here's what I see:
>
> postgres@salmo:$ postgres -D /usr/local/pgsql/data &
> [1] 16044
> postgres@salmo:/var/log/httpd$ LOG:  could not open directory
> "/share/timezone": No such file or directory
> LOG:  could not open directory "/share/timezone": No such file or
> directory
> LOG:  could not open directory "/share/timezone": No such file or
> directory
> LOG:  could not open directory "/share/timezone": No such file or
> directory
> LOG:  could not open directory "/share/timezone": No such file or
> directory
> WARNING:  could not open directory "/share/timezonesets": No such file or
> directory
> HINT:  This may indicate an incomplete PostgreSQL installation, or
> that the
> file "/bin/postgres" has been moved away from its proper location.
> FATAL:  invalid value for parameter "timezone_abbreviations": "Default"
>
>   The files and directories are in /usr/local/pgsql/share:

if you built and installed postgres in /usr/local/pgsql, then all those
errors should have referred to /usr/local/pgsql/share/timezone

is there a different postgres binary in /bin/postgres ?

execute...

postgres@salmo:$ /usr/local/pgsql/bin/pg_config

and reply to the list with the output...

also,

postgres@salmo:$ which postgres



Re: Problems With -9.0.1

From
Rich Shepard
Date:
On Mon, 20 Dec 2010, Adrian Klaver wrote:

> As I remember it there was more than one version of Postgres on this
> machine. Are you sure you are using the correct postgres binary? While I
> am it is there a reason you are not using the system start scripts or
> pg_ctl:)?

Adrian,

   There _was_ a library not removed when I removed -8.3.3, but that's been
gone a while. This had been running until yesterday when I could not log in
to my accounting software that uses postgres as the back end.

   The reason I don't use /etc/rc.d/rc.postgresql is that it doesn't work.
While it supposedly su's to user postgres, it actually does not do so. Ergo,
nothing starts.

   Looking for running postgres processes I find none:

[root@salmo ~]# ps ax | grep postgres
17168 pts/2    S+     0:00 grep postgres

   And when I try to start the postmaster as user postgres it fails (as
reported in my original message):

postgres@salmo:/home/rshepard$ LOG:  could not open directory
"/share/timezone": No such file or directory
LOG:  could not open directory "/share/timezone": No such file or directory
LOG:  could not open directory "/share/timezone": No such file or directory
LOG:  could not open directory "/share/timezone": No such file or directory
LOG:  could not open directory "/share/timezone": No such file or directory
WARNING:  could not open directory "/share/timezonesets": No such file or
directory
HINT:  This may indicate an incomplete PostgreSQL installation, or that the
file "/bin/postgres" has been moved away from its proper location.
FATAL:  invalid value for parameter "timezone_abbreviations": "Default"

[1]+  Exit 1                  postgres -D /usr/local/pgsql/data

   This has not happened before.

Rich

Re: Problems With -9.0.1 [SOLVED]

From
Rich Shepard
Date:
On Mon, 20 Dec 2010, John R Pierce wrote:

> if you built and installed postgres in /usr/local/pgsql, then all those
> errors should have referred to /usr/local/pgsql/share/timezone

John,

   I assumed the reference to share/ was relative to /usr/local/pgsql/

> is there a different postgres binary in /bin/postgres ?

   Aha! I had a copy of /usr/local/pgsql/bin/postgres in /bin rather than a
soft link. I just fixed that and postgres starts. Still cannot log in to my
accounting software, even after restarting httpd. That brings me back to my
original problem: server error that won't allow me to log in.

   There's a problem with the perl DBD::Pg module that prevents it from
loading. Sigh. I'll focus on that now.

Thanks,

Rich

Re: Problems With -9.0.1

From
Adrian Klaver
Date:
On Monday 20 December 2010 7:12:52 pm Rich Shepard wrote:
> On Mon, 20 Dec 2010, Adrian Klaver wrote:
> > As I remember it there was more than one version of Postgres on this
> > machine. Are you sure you are using the correct postgres binary? While I
> > am it is there a reason you are not using the system start scripts or
> > pg_ctl:)?
>
> Adrian,
>
>    There _was_ a library not removed when I removed -8.3.3, but that's been
> gone a while. This had been running until yesterday when I could not log in
> to my accounting software that uses postgres as the back end.
>
>    The reason I don't use /etc/rc.d/rc.postgresql is that it doesn't work.
> While it supposedly su's to user postgres, it actually does not do so.
> Ergo, nothing starts.

Well that would be a problem. Still, using pg_ctl would be an improvement. From
the docs (http://www.postgresql.org/docs/9.0/interactive/app-pg-ctl.html):
"pg_ctl is a utility for initializing a PostgreSQL database cluster, starting,
stopping, or restarting the PostgreSQL backend server (postgres), or displaying
the status of a running server. Although the server can be started manually,
pg_ctl encapsulates tasks such as redirecting log output and properly detaching
from the terminal and process group. It also provides convenient options for
controlled shutdown."

>
>    Looking for running postgres processes I find none:
>
> [root@salmo ~]# ps ax | grep postgres
> 17168 pts/2    S+     0:00 grep postgres

The issue is not so much another running postgres but more than one in your PATH
and you picking up the wrong one. A locate or find would help here.

Also per Johns post you should run:
/usr/local/pgsql/bin/pg_config

That may help untangle this.


>
>    And when I try to start the postmaster as user postgres it fails (as




>
> Rich



--
Adrian Klaver
adrian.klaver@gmail.com

Re: Problems With -9.0.1

From
Rich Shepard
Date:
On Mon, 20 Dec 2010, Adrian Klaver wrote:

> Well that would be a problem. Still, using pg_ctl would be an improvement.
> From the docs
> (http://www.postgresql.org/docs/9.0/interactive/app-pg-ctl.html): "pg_ctl
> is a utility for initializing a PostgreSQL database cluster, starting,
> stopping, or restarting the PostgreSQL backend server (postgres), or
> displaying the status of a running server. Although the server can be
> started manually, pg_ctl encapsulates tasks such as redirecting log output
> and properly detaching from the terminal and process group. It also
> provides convenient options for controlled shutdown."

   The script stops the running postgres. Since the only time I should need
to re-start it is after a re-boot (a distribution upgrade with a new
kernel), starting manually is not a big deal:

postgres@salmo:/home/rshepard$ postgres -D /usr/local/pgsql/data &
[1] 17318
postgres@salmo:/home/rshepard$ LOG:  database system was shut down at
2010-12-20 19:04:58 PST
LOG:  database system is ready to accept connections
LOG:  autovacuum launcher started

[rshepard@salmo ~]$ ps ax | grep postgres
17318 pts/1    S      0:00 postgres -D /usr/local/pgsql/data
17320 ?        Ss     0:00 postgres: writer process
17321 ?        Ss     0:00 postgres: wal writer process
17322 ?        Ss     0:00 postgres: autovacuum launcher process
17323 ?        Ss     0:00 postgres: stats collector process
17405 pts/1    S+     0:00 grep postgres

   All processes running.

Thanks, Adrian,

Rich