Обсуждение: initdb fails on ia64 SMP system

Hello PostgreSQL people,

Thanks for making PostgreSQL, it's great. Here's a bug report.

Thanks
Ian

Your name               : Ian Grant
Your email address      : ian.grant@cl.cam.ac.uk

System Configuration
---------------------
  Architecture (example: Intel Pentium)         : ia64, 4 way SMP

  Operating System (example: Linux 2.0.26 ELF)  : Linux 2.4.3 SMP kernel

  PostgreSQL version (example: PostgreSQL-7.2.1):   PostgreSQL-7.2.1

  Compiler used (example:  gcc 2.95.2)  : 2.96 20000731 (Red Hat Linux 7.1
2.96-85)

Please enter a FULL description of your problem:
-------------------------------------------------

Running "initdb -D /data/path" fails. With debug enabled (-d switch) I get a
trace which ends:

DEBUG:  inserting column 10 value 'byteain'
DEBUG:  Typ == NULL, typeindex = 7
DEBUG:   -> byteain
DEBUG:  inserted
DEBUG:  inserting column 11 value 'byteaout'
DEBUG:  Typ == NULL, typeindex = 7
ERROR:  Relation "pg_class" does not exist

initdb failed.
Removing /local/scratch/ig206/postgresql_data.

Running the same command again causes it to fail somewhere different, e.g.:

DEBUG:  inserting column 9 value '0'
DEBUG:  Typ == NULL, typeindex = 9
DEBUG:   -> 0
DEBUG:  inserted
DEBUG:  inserting column 10 value 'cidin'
DEBUG:  Typ == NULL, typeindex = 7
DEBUG:   -> cidin
DEBUG:  inserted
DEBUG:  inserting column 11 value 'cidout'
DEBUG:  Typ == NULL, typeindex = 7
ERROR:  Relation "pg_class" does not exist

initdb failed.
Removing /local/scratch/ig206/postgresql_data.

It's not always pg_class which doesn't exist, e.g.:

DEBUG:  inserting column 10 value 'boolin'
DEBUG:  Typ == NULL, typeindex = 7
DEBUG:   -> boolin
DEBUG:  inserted
DEBUG:  inserting column 11 value 'boolout'
DEBUG:  Typ == NULL, typeindex = 7
ERROR:  Relation "pg_proc" does not exist


Please describe a way to repeat the problem.   Please try to provide a
concise reproducible example, if at all possible:
-----------------------------------------------------------------------

Build postgresql 7.2.1 on an SMP ia64 machine and run initdb?

If you know how this problem might be fixed, list the solution below:
----------------------------------------------------------------------

Sorry, I have no idea at all.

--
Ian Grant, Computer Lab., William Gates Building, JJ Thomson Ave., Cambridge
Phone: +44 1223 334420

--
Ian Grant, Computer Lab., William Gates Building, JJ Thomson Ave., Cambridge
Phone: +44 1223 334420

Ian Grant <Ian.Grant@cl.cam.ac.uk> writes:
>   Architecture (example: Intel Pentium)         : ia64, 4 way SMP
>   Operating System (example: Linux 2.0.26 ELF)  : Linux 2.4.3 SMP kernel

Does anything else work for you on that box?  I seem to recall that
kernels before about 2.4.9 had serious SMP bugs.

> Running the same command again causes it to fail somewhere different, e.g.:

In that case I don't think it's our bug.  initdb is quite deterministic,
at least when running on a deterministic platform ;-)

            regards, tom lane

> Ian Grant <Ian.Grant@cl.cam.ac.uk> writes:
> >   Architecture (example: Intel Pentium)         : ia64, 4 way SMP
> >   Operating System (example: Linux 2.0.26 ELF)  : Linux 2.4.3 SMP kernel
>
> Does anything else work for you on that box?  I seem to recall that
> kernels before about 2.4.9 had serious SMP bugs.

No, not much that I've tried, but its owners don't seem to notice.

> > Running the same command again causes it to fail somewhere different, e.g.:
>
> In that case I don't think it's our bug.  initdb is quite deterministic,
> at least when running on a deterministic platform ;-)

I didn't know this. I will try 2.4.9 and get back to you.

Many thanks,
Ian
--
Ian Grant, Computer Lab., William Gates Building, JJ Thomson Ave., Cambridge
Phone: +44 1223 334420

> Ian Grant <Ian.Grant@cl.cam.ac.uk> writes:
> >   Architecture (example: Intel Pentium)         : ia64, 4 way SMP
> >   Operating System (example: Linux 2.0.26 ELF)  : Linux 2.4.3 SMP kernel
>
> Does anything else work for you on that box?  I seem to recall that
> kernels before about 2.4.9 had serious SMP bugs.

Thanks. That was it.

Best wishes
Ian
--
Ian Grant, Computer Lab., William Gates Building, JJ Thomson Ave., Cambridge
Phone: +44 1223 334420