Thread

1. runtest bug and bigcheck problems

   Rick Touchette <rick.touchette@nwa.com> — 2001-04-06T18:25:27Z

   Some background:
     Version: 	postgresql-7.0.3  installed in nonstandard location
   (/home/nssrta/pgsql)
     OS:		 Solaris 2.6 (patch level unknown, but definitely not current)
     Running under the installed user id
   
   Item #1 Bug in runtest:
   In order to get the runtest regression test to work I had to change line 100
   in regress.sh
     from:
        createlang $HOSTLOC plpgsql regression
     to:
        createlang -L /home/n33rta/pgsql/lib $HOSTLOC plpgsql regression
        (I had to specify the lib location)
   
   There should be a way to get this value from the Makefile.global which is
   included in the GNUmakefile, But GNUmakefile is running regress.sh in a
   shell. Sorry, I'm not very knowledgeable with gmake and therefore I haven't
   a eloquent solution to suggest.
   
   Item #2
   Before install the runcheck test worked fine but when I tried the bigcheck
   test it hung.
   that is the test ran fine until the big (whatever) test and never finished
   (4+ hours)
   I did notice a defunct process from which had a common parent to the
   postmaster process.
   
   Any ideas??
   
   
   Thanks,
   Rick Touchette
   
   
   

2. Re: runtest bug and bigcheck problems

   Tom Lane <tgl@sss.pgh.pa.us> — 2001-04-13T16:06:26Z

   "Rick Touchette" <rick.touchette@nwa.com> writes:
   > In order to get the runtest regression test to work I had to change line 100
   > in regress.sh
   >   from:
   >      createlang $HOSTLOC plpgsql regression
   >   to:
   >      createlang -L /home/n33rta/pgsql/lib $HOSTLOC plpgsql regression
   >      (I had to specify the lib location)
   
   This is fixed in 7.1.
   
   > Item #2
   > Before install the runcheck test worked fine but when I tried the bigcheck
   > test it hung.
   > that is the test ran fine until the big (whatever) test and never finished
   > (4+ hours)
   
   If you got through the regular test then it's highly unlikely that
   bigcheck would really fail --- it's simply a repeat of the regular
   numeric test with (many) more digit positions.  I think you just didn't
   wait long enough.  The big numeric test does take a ridiculously long
   time...
   
   			regards, tom lane