First, please accept my apologies and advise me accordingly if I am posting
this to the wrong list. I am at wit's end trying to solve this problem.
The platform I am working with is Solaris 10, with some servers being
standalone, others being zones. They are divided into three environments:
development (one zone), QA (one zone) and production (two standalone).
We are, by company convention, using /opt as the installation location,
specifically, any given build of PostgreSQL goes into /opt/postgres/x.y.z, with
x.y.z being filled in with the current version number (presently 8.3.7). A
symbolic link then points from /opt/postgres/current to /opt/postgres/x.y.z,
and another symlink points from /usr/local/pgsql to /opt/postgres/current.
Without getting into the whys of all of these, please accept that these facts
are unchangeable.
On the development zone, there are no problems at all. I can do this:
./configure --prefix=/opt/postgres/8.3.7 && make -j16 && make
-j16 check && make -j16 install
. . . and it just works.
On the QA zone or Production servers, however the same command (adjusted for
the number of processors) fails at 'make check' thus:
---8<-cut----------------------------------------------------------------------
./pg_regress --temp-install=./tmp_check --top-builddir=../../.. --srcdir=/usr/us
ers/postgres/build/postgresql-8.3.7/src/test/regress --temp-port=55432 --schedul
e=./parallel_schedule --multibyte=SQL_ASCII --load-language=plpgsql
============== creating temporary installation ==============
============== initializing database system ==============
============== starting postmaster ==============
Killed
Killed
*
Killed
Killed
pg_regress: postmaster did not respond within 60 seconds
Examine ./log/postmaster.log for the reason
make[2]: *** [check] Error 2
make[2]: Leaving directory `/usr/users/postgres/build/postgresql-8.3.7/src/test/
regress'
make[1]: *** [check] Error 2
make[1]: Leaving directory `/usr/users/postgres/build/postgresql-8.3.7/src/test'
make: *** [check] Error 2
---8<-cut----------------------------------------------------------------------
The "Killed" message repeats 60 times at what appear to be approximately
one-second intervals.
I have tried the following:
* configure and make on dev machine; check on QA -- check fails as above.
* configure and make on QA machine; check on dev -- check succeeds.
* configure, make, check on prod 1 or prod 2 machines -- check fails on either
machine.
* configure, make on prod 1 or prod 2, move to dev and check -- check succeeds.
One last tidbit: If I leave off the --prefix= option on the configure, the
tests run just fine on all machines.
Although I *could* build and check on the dev machine, and then just 'make
install' on the other machines, in my opinion, I feel that I must know what is
going wrong here. I need to know why it is b0rked, so that, should this
problem manifest elsewhere, it can be fixed. Should the UNIX admins
accidentally make a change to the dev zone that causes it to start this
behaviour, then I am completely sunk. Should they upgrade the dev server and
cause this problem to manifest, then I am likewise sunk. Getting PostgreSQL
into this organization is a delicate issue, and, as an open-source advocate, I
want to keep it here. This necessarily means satisfying bureaucratic
requirements that I am not empowered to change.
Many thanks in advance for any assistance that can be provided.
**********************************************************************
The information in this email is confidential and may be legally
privileged against disclosure other than to the intended recipient.
It is intended solely for the addressee. Access to this email by
anyone else is unauthorized.
If you are not the intended recipient, any disclosure, copying,
distribution or any action taken or omitted to be taken in reliance
on it, is prohibited and may be unlawful. Please immediately
delete this message and inform the sender of this error.
**********************************************************************
