Thread: Problem with regression test

My apologies if this is the wrong list - my message was bounced from the
address in the documentation.

Hi,

Tried to set up postgreSQL v6.5 on a:

486DX66
32Mb RAM
4.3 Gb HDD  (less than 1Gb used)
RedHat Linux 5.2 (2.0.36)

Uncompressing the file left me with a 'postgresql-6.5' directory which
had to be included in the path (this isn't shown in the instructions).
Make succeeded, and I was able to run the 'createdb' example in the
instructions.
Regression test failed at this point:

...
select_views .. ok
alter_table .. ERROR:  Unable to locate type name 'dt' in catalog
ERROR:  Relation 'tmp' does not have attribute 'k'
ERROR:  Unable to locate type name 'dt' in catalog
ERROR:  Relation 'tmp' does not have attribute 'k'
ok
portals_p2 .. ok
rules .. failed
limit .. ok
plpgsql .. ERROR:  Cannot insert a duplicate key into a unique index
ERROR:  WS.not.there         does not exists
ERROR:  illegal backlink beginning with XX
ERROR:  PS.not.there         does not exists
ERROR:  illegal slotlink beginning with XX
ERROR:  Cannot insert a duplicate key into a unique index
ERROR:  no manual manipulation of HSlot
ERROR:  no manual manipulation of HSlot
ERROR:  system "notthere" does not exist
ERROR:  IFace slotname "IF.orion.ethernet_interface_name_too_long" too
long (20 char max)
ok
temp .. ERROR:  temptest: Table does not exist.
ok
numeric .. DEBUG:  --Relation num_exp_add--
DEBUG:  Pages 2: Changed 0, Reapped 0, Empty 0, New 0; Tup 100: Vac 0,
Keep/VTL 0/0, Crash 0, UnUsed 0, MinLen 50, MaxLen 118; Re-using:
Free/Avail. Space 0/0; EndEmpty/Avail. Pages 0/0. Elapsed 0/0 sec.
DEBUG:  Index num_exp_add_idx: Pages 2; Tuples 100. Elapsed 0/0 sec.
DEBUG:  --Relation num_exp_sub--
DEBUG:  Pages 2: Changed 0, Reapped 0, Empty 0, New 0; Tup 100: Vac 0,
Keep/VTL 0/0, Crash 0, UnUsed 0, MinLen 50, MaxLen 119; Re-using:
Free/Avail. Space 0/0; EndEmpty/Avail. Pages 0/0. Elapsed 0/0 sec.
DEBUG:  Index num_exp_sub_idx: Pages 2; Tuples 100. Elapsed 0/0 sec.
DEBUG:  --Relation num_exp_div--
DEBUG:  Pages 2: Changed 0, Reapped 0, Empty 0, New 0; Tup 100: Vac 0,
Keep/VTL 0/0, Crash 0, UnUsed 0, MinLen 50, MaxLen 118; Re-using:
Free/Avail. Space 0/0; EndEmpty/Avail. Pages 0/0. Elapsed 0/0 sec.
DEBUG:  Index num_exp_div_idx: Pages 2; Tuples 100. Elapsed 0/0 sec.
DEBUG:  --Relation num_exp_mul--
DEBUG:  Pages 2: Changed 0, Reapped 0, Empty 0, New 0; Tup 100: Vac 0,
Keep/VTL 0/0, Crash 0, UnUsed 0, MinLen 50, MaxLen 137; Re-using:
Free/Avail. Space 0/0; EndEmpty/Avail. Pages 0/0. Elapsed 0/0 sec.
DEBUG:  Index num_exp_mul_idx: Pages 2; Tuples 100. Elapsed 0/0 sec.
DEBUG:  --Relation num_exp_sqrt--
DEBUG:  Pages 1: Changed 0, Reapped 0, Empty 0, New 0; Tup 10: Vac 0,
Keep/VTL 0/0, Crash 0, UnUsed 0, MinLen 46, MaxLen 106; Re-using:
Free/Avail. Space 0/0; EndEmpty/Avail. Pages 0/0. Elapsed 0/0 sec.
DEBUG:  Index num_exp_sqrt_idx: Pages 2; Tuples 10. Elapsed 0/0 sec.
DEBUG:  --Relation num_exp_ln--
DEBUG:  Pages 1: Changed 0, Reapped 0, Empty 0, New 0; Tup 10: Vac 0,
Keep/VTL 0/0, Crash 0, UnUsed 0, MinLen 46, MaxLen 97; Re-using:
Free/Avail. Space 0/0; EndEmpty/Avail. Pages 0/0. Elapsed 0/0 sec.
DEBUG:  Index num_exp_ln_idx: Pages 2; Tuples 10. Elapsed 0/0 sec.
DEBUG:  --Relation num_exp_log10--
DEBUG:  Pages 1: Changed 0, Reapped 0, Empty 0, New 0; Tup 10: Vac 0,
Keep/VTL 0/0, Crash 0, UnUsed 0, MinLen 46, MaxLen 97; Re-using:
Free/Avail. Space 0/0; EndEmpty/Avail. Pages 0/0. Elapsed 0/0 sec.
DEBUG:  Index num_exp_log10_idx: Pages 2; Tuples 10. Elapsed 0/0 sec.
DEBUG:  --Relation num_exp_power_10_ln--
DEBUG:  Pages 1: Changed 0, Reapped 0, Empty 0, New 0; Tup 10: Vac 0,
Keep/VTL 0/0, Crash 0, UnUsed 0, MinLen 46, MaxLen 139; Re-using:
Free/Avail. Space 0/0; EndEmpty/Avail. Pages 0/0. Elapsed 0/0 sec.
DEBUG:  Index num_exp_power_10_ln_idx: Pages 2; Tuples 10. Elapsed 0/0
sec.

at this point, the program apparently hung (using over 90% CPU) for
several minutes, until I killed the 'regression' process as root.  The
xterm then displayed:

failed
ACTUAL RESULTS OF REGRESSION TEST ARE NOW IN FILE regress.out
...

I can still run commands like 'createdb', etc.

Hopefully, this is a simple-minded error on my part.  Thanks for any
help you can offer.

Jim

> at this point, the program apparently hung (using over 90% CPU) for
> several minutes, until I killed the 'regression' process as root.  The
> xterm then displayed:
> failed
> ACTUAL RESULTS OF REGRESSION TEST ARE NOW IN FILE regress.out
> ...
> I can still run commands like 'createdb', etc.
> Hopefully, this is a simple-minded error on my part.  Thanks for any
> help you can offer.

I'm guessing that you ran out of some sort of resource on your machine
(e.g. swap space?). With a 486DX66 and 32Mb RAM you might be pushing
your luck on the regression tests, though others may have had success
so don't take my word for it. 

The regression test reference machine is a PPro RH5.2 system, so if
your installation went OK (it looks like it did) don't worry about the
regression test results. btw, those "ERROR" messages you see are,
afaict, all normal; parts of the regression test explicitly test edge
cases to make sure we handle errors correctly.

Good luck.
                    - Thomas

-- 
Thomas Lockhart                lockhart@alumni.caltech.edu
South Pasadena, California

Jim Lemon <Jim.Lemon@uts.EDU.AU> writes:
> Regression test failed at this point:
> ...
> select_views .. ok
> alter_table .. ERROR:  Unable to locate type name 'dt' in catalog
> ERROR:  Relation 'tmp' does not have attribute 'k'
> ERROR:  Unable to locate type name 'dt' in catalog
> ERROR:  Relation 'tmp' does not have attribute 'k'
> ok

These "ERROR" messages are in fact normal log output during the
regression test ... but they shouldn't be appearing on your screen!
I think that when you started the postmaster, you forgot to redirect
its stdout and stderr into a log file per the instructions.  The
regression tests will usually generate something like 50K of log output
(since they deliberately provoke a lot of errors...)

> rules .. failed

That's a little more disturbing, but you'll need to look at the
regression.diffs output to see if the difference is actually meaningful.
I suspect there are still a few queries in the rules test wherein the
order in which tuples are displayed is platform-dependent ... don't
worry about it if the differences turn out to look like that.

> at this point, the program apparently hung (using over 90% CPU) for
> several minutes, until I killed the 'regression' process as root.

Patience, my boy ... the last 'numeric' test is pretty slow, but it's
not hung.
        regards, tom lane

> > at this point, the program apparently hung (using over 90% CPU) for
> > several minutes, until I killed the 'regression' process as root.
> Patience, my boy ... the last 'numeric' test is pretty slow, but it's
> not hung.

Good point! I'd forgotten what this would be like on a 486 machine :)
                 - Thomas

-- 
Thomas Lockhart                lockhart@alumni.caltech.edu
South Pasadena, California