Thread: grant still broken

grant still broken

From
Peter T Mount
Date:
Well, I've rebuilt postgres again, using a fresh cvs snapshot, and grant
is still broken.

Infact, it gets worse. Once you try to use the grant statement, then any
other connection that was active at that time - even on different
databases, will die with the same message, hinting it may be something in
shared memory that's breaking.

Heres the output of psql, first as user postgres:

[postgres@maidast pgsql]$ psql -h localhost db1
Welcome to the POSTGRESQL interactive sql monitor:
  Please read the file COPYRIGHT for copyright terms of POSTGRESQL

   type \? for help on slash commands
   type \q to quit
   type \g or terminate with semicolon to execute query
 You are currently connected to the database: db1

db1=> create table a (id int4,url text);
CREATE
db1=> insert into a values (1,'http://www.demon.co.uk/finder');
INSERT 18250 1
db1=> grant all on a to public;
PQexec() -- Request was sent to backend, but backend closed the channel
before responding.
        This probably means the backend terminated abnormally before or
while processing the request.

And as user pmount:

[pmount@maidast pmount]$ psql -h localhost
Welcome to the POSTGRESQL interactive sql monitor:
  Please read the file COPYRIGHT for copyright terms of POSTGRESQL

   type \? for help on slash commands
   type \q to quit
   type \g or terminate with semicolon to execute query
 You are currently connected to the database: pmount

pmount=> create table a (id int4,url text);
CREATE
pmount=> insert into a values (1,'http://www.demon.co.uk/finder');
INSERT 18314 1
pmount=> select * from a;
id|url
--+-----------------------------
 1|http://www.demon.co.uk/finder
(1 row)

pmount=> select * from a;
PQexec() -- Request was sent to backend, but backend closed the channel
before responding.
        This probably means the backend terminated abnormally before or
while processing the request.

NB: the select done as pmount was after the postgres user tried to grant
rights, while the other statements were before.

--
Peter T Mount  petermount@earthling.net or pmount@maidast.demon.co.uk
Main Homepage: http://www.demon.co.uk/finder
Work Homepage: http://www.maidstone.gov.uk Work EMail: peter@maidstone.gov.uk


Re: [HACKERS] grant still broken

From
"Thomas G. Lockhart"
Date:
> Well, I've rebuilt postgres again, using a fresh cvs snapshot, and grant
> is still broken.

I see the same behavior here, so it's not just you :)

                                      - Tom


Re: [HACKERS] grant still broken

From
Bruce Momjian
Date:
>
> Well, I've rebuilt postgres again, using a fresh cvs snapshot, and grant
> is still broken.
>
> Infact, it gets worse. Once you try to use the grant statement, then any
> other connection that was active at that time - even on different
> databases, will die with the same message, hinting it may be something in
> shared memory that's breaking.
>
> Heres the output of psql, first as user postgres:
>
> [postgres@maidast pgsql]$ psql -h localhost db1
> Welcome to the POSTGRESQL interactive sql monitor:
>   Please read the file COPYRIGHT for copyright terms of POSTGRESQL
>
>    type \? for help on slash commands
>    type \q to quit
>    type \g or terminate with semicolon to execute query
>  You are currently connected to the database: db1
>
> db1=> create table a (id int4,url text);
> CREATE
> db1=> insert into a values (1,'http://www.demon.co.uk/finder');
> INSERT 18250 1
> db1=> grant all on a to public;
> PQexec() -- Request was sent to backend, but backend closed the channel
> before responding.
>         This probably means the backend terminated abnormally before or
> while processing the request.

Gee, still works fine here under bsd/os 3.0.

    #$ sql test
    Welcome to the POSTGRESQL interactive sql monitor:
      Please read the file COPYRIGHT for copyright terms of POSTGRESQL

       type \? for help on slash commands
       type \q to quit
       type \g or terminate with semicolon to execute query
     You are currently connected to the database: test

    test=> create table a (id int4,url text);
    CREATE
    test=> insert into a values (1,'http://www.demon.co.uk/finder');
    INSERT 143530 1
    test=> grant all on a to public;
    CHANGE
    test=>


--
Bruce Momjian
maillist@candle.pha.pa.us

Re: [HACKERS] grant still broken

From
The Hermit Hacker
Date:
On Tue, 13 Jan 1998, Bruce Momjian wrote:

> Gee, still works fine here under bsd/os 3.0.
>
>     #$ sql test
>     Welcome to the POSTGRESQL interactive sql monitor:
>       Please read the file COPYRIGHT for copyright terms of POSTGRESQL
>
>        type \? for help on slash commands
>        type \q to quit
>        type \g or terminate with semicolon to execute query
>      You are currently connected to the database: test
>
>     test=> create table a (id int4,url text);
>     CREATE
>     test=> insert into a values (1,'http://www.demon.co.uk/finder');
>     INSERT 143530 1
>     test=> grant all on a to public;
>     CHANGE
>     test=>

FreeBSD 3.0-CURRENT as of yesterday:

scrappy=> create table a (id int4,url text);
CREATE
scrappy=> insert into a values (1,'http://www.demon.co.uk/finder');
INSERT 143210 1
scrappy=> grant all on a to public;
CHANGE
scrappy=> \z

Database    = scrappy
 +------------------+----------------------------------------------------+
 |  Relation        |             Grant/Revoke Permissions               |
 +------------------+----------------------------------------------------+
 | a                | {"=arwR"}                                          |
 | testtable        |                                                    |
 +------------------+----------------------------------------------------+
scrappy=>

Marc G. Fournier
Systems Administrator @ hub.org
primary: scrappy@hub.org           secondary: scrappy@{freebsd|postgresql}.org


Re: [HACKERS] grant still broken

From
Peter T Mount
Date:
On Wed, 14 Jan 1998, Thomas G. Lockhart wrote:

> > Well, I've rebuilt postgres again, using a fresh cvs snapshot, and grant
> > is still broken.
>
> I see the same behavior here, so it's not just you :)

Tom, what platform are you running it on? It seems the others have no
problems with BSD.

I'm using Linux.

--
Peter T Mount  petermount@earthling.net or pmount@maidast.demon.co.uk
Main Homepage: http://www.demon.co.uk/finder
Work Homepage: http://www.maidstone.gov.uk Work EMail: peter@maidstone.gov.uk


Re: [HACKERS] grant still broken

From
James Hughes
Date:

On Wed, 14 Jan 1998, Peter T Mount wrote:

> On Wed, 14 Jan 1998, Thomas G. Lockhart wrote:
>
> > > Well, I've rebuilt postgres again, using a fresh cvs snapshot, and grant
> > > is still broken.
> >
> > I see the same behavior here, so it's not just you :)
>
> Tom, what platform are you running it on? It seems the others have no
> problems with BSD.
>
> I'm using Linux.
>

I'm running Linux and I'm getting the current cvs tree now. I will try to
reproduce the problem on my system and track it down.

I am still running 6.2.0. Anything I need to watch out for when running
the latest?



-James



Re: [HACKERS] grant still broken

From
"Thomas G. Lockhart"
Date:
> > > Well, I've rebuilt postgres again, using a fresh cvs snapshot, and grant
> > > is still broken.
> >
> > I see the same behavior here, so it's not just you :)
>
> Tom, what platform are you running it on? It seems the others have no
> problems with BSD.
>
> I'm using Linux.

Me too. Shouldn't be platform-specific, but maybe it will be clear once the
solution is found. There was a reported problem with "group by" sorting which I
could reproduce but which other platforms could not; it went away after a month
or so. Maybe we'll get lucky on this one too :/

                                            - Tom


Re: [HACKERS] grant still broken

From
"Thomas G. Lockhart"
Date:
> > > > Well, I've rebuilt postgres again, using a fresh cvs snapshot, and grant
> > > > is still broken.
> > >
> > > I see the same behavior here, so it's not just you :)
> >
> > Tom, what platform are you running it on? It seems the others have no
> > problems with BSD.
> >
> > I'm using Linux.
>
> I'm running Linux and I'm getting the current cvs tree now. I will try to
> reproduce the problem on my system and track it down.
>
> I am still running 6.2.0. Anything I need to watch out for when running
> the latest?

Just watch out for better performance :)

Oh, there is a problem with varchar() at the moment.


Re: [HACKERS] grant still broken

From
James Hughes
Date:

On Wed, 14 Jan 1998, Thomas G. Lockhart wrote:

> > > > > Well, I've rebuilt postgres again, using a fresh cvs snapshot, and grant
> > > > > is still broken.
> > > >
> > > > I see the same behavior here, so it's not just you :)
> > >
> > > Tom, what platform are you running it on? It seems the others have no
> > > problems with BSD.
> > >
> > > I'm using Linux.
> >
> > I'm running Linux and I'm getting the current cvs tree now. I will try to
> > reproduce the problem on my system and track it down.
> >
> > I am still running 6.2.0. Anything I need to watch out for when running
> > the latest?
>
> Just watch out for better performance :)
>
> Oh, there is a problem with varchar() at the moment.
>

I've (quickly) compiled and installed this mornings cvs and am getting some
odd behavior. I have a death in the family, so I'm not going to be able to
sort it all out before this weekend.

I get segfaults and authentication problems when trying to run perl scripts.
Also, I'm not sure I have a handle on the new acl stuff.


`Anyways', hopefully I can get everything worked out and compile a debug
version to poke through with gdb this weekend.


-James