Skip to main content

32-bit PostgreSQL Compilation On 64-bit CentOS 6.x

I am sure that, most of you aware of this. But, for me it's the first time, I accomplished it.

As one of my assigned tasks to build a 32-bit instance of postgresql on 64-bit machine, I have followed the below approach. I hope, it will be helpful to others as well, if you got any problems.

As an initial step on this task, I have tried to build a sample "c" program using "gcc -m32". Once, I resolved this, I moved to compile the PostgreSQL 9.0.
[root@localhost Desktop]# gcc -m32 -o test test.c
In file included from /usr/include/features.h:385,
                 from /usr/include/stdio.h:28,
                 from test.c:1:
/usr/include/gnu/stubs.h:7:27: error: gnu/stubs-32.h: No such file or directory
To resolve the above issue, I have installed the 32-bit glibc-devel package through yum.
yum -y install glibc-devel.i686 glibc-devel
Again, I have tried to run the same command.
[root@localhost Desktop]# gcc -m32 -o test test.c
/usr/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-redhat-linux/4.4.6/libgcc_s.so when searching for -lgcc_s
/usr/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-redhat-linux/4.4.6/libgcc_s.so when searching for -lgcc_s
/usr/bin/ld: cannot find -lgcc_s
collect2: ld returned 1 exit status
Now, I got a different error message, and tried to install the 32-bit libgcc.
[root@localhost Desktop]# yum install libgcc-*.i686
--> Running transaction check
---> Package libgcc.i686 0:4.4.7-4.el6 will be installed
--> Finished Dependency Resolution
Error: Protected multilib versions: libgcc-4.4.7-4.el6.i686 != libgcc-4.4.6-3.el6.x86_64
As it's complaining, we have the old version of x86_64, when compared the new one. Hence, I have tried to update the existing x86_64.
[root@localhost Desktop]# yum update libgcc-4.4.6-3.el6.x86_64
---> Package libgcc.x86_64 0:4.4.6-3.el6 will be updated
---> Package libgcc.x86_64 0:4.4.7-4.el6 will be an update
--> Finished Dependency Resolution
Once, it's updated the given library, I have again tried to install the libgcc-*.i686.
[root@localhost Desktop]# yum install libgcc-*.i686
Resolving Dependencies
--> Running transaction check
---> Package libgcc.i686 0:4.4.7-4.el6 will be installed
--> Finished Dependency Resolution
Now, I am trying to run the same "gcc -m32" command to check for any further issues.
[root@localhost Desktop]# gcc -m32 -o test test.c
[root@localhost Desktop]# ./test
Hello World
It looks, the sample "c program" is working fine, as a 32-bit application. Now, i am moving to PostgreSQL 9.0. As per my observation I have updated, installed the below components for the PostgreSQL.
yum update readline-6.0-3.el6.x86_64
yum install *readline*i686
yum update zlib-1.2.3-27.el6.x86_64
yum install *zlib*i686
Once, I got the all required 32-bit libraries, I have tried to compile the postgresql as below.
[root@localhost build]# CFLAGS=-m32 LDFLAGS=-m32 CXXFLAGS=-m32 ./configure --prefix=/opt/PostgreSQL/build
[root@localhost build]# make -j 4; make install;
....
make[1]: Leaving directory `/root/Downloads/postgresql-9.0.17/config'
PostgreSQL installation complete.
It seems, postgresql has built successfully on 64-bit machine as a 32-bit application.

Checking for the confirmation from OS, PG.
postgres=# SELECT version();
                                                      version                                                      
-------------------------------------------------------------------------------------------------------------------
 PostgreSQL 9.0.17 on x86_64-unknown-linux-gnu, compiled by GCC gcc (GCC) 4.4.6 20110731 (Red Hat 4.4.6-3), 32-bit
(1 row)

[root@localhost build]# file /sbin/init
/sbin/init: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.18, stripped
And, at the end, I have the postgresql as 32-bit application in 64-bit machine.

Thank you for reading, and please comment on this, if you have any questions.


--Dinesh Kumar

Comments

  1. This is clever and all, but you've left out the part that explains why someone would want to do this in the first place. Any clues?

    ReplyDelete
  2. Thanks David.

    Being frank with you, i do not know, why someone need this to implement in their architecture. I have just tried to accomplish the task, what i got from my higher levels. I thought, this post would be helpful to others if they face the similar kind of situation.

    ReplyDelete
  3. What my assumption is, the end customer is trying their native 32-bit postgresql in a 64-bit machine. May be that's reason, he want all 32-bit libraries of PostgreSQL in 64-bit machine.

    ReplyDelete
  4. Hey, I want to thank you for posting what you discovered. There's still lots of confusion around 32- versus 64-bit and some vendors still require 32-bit tools. Your post helped me past a major stumbling block with one such vendor. Thanks!

    ReplyDelete
  5. Ditto. I am working with 32 bit libraries on a 64 bit Centos machine, and this helped me solve my problem. Thanks!

    ReplyDelete
  6. My build of a 32bit app worked nicely on centos5 through centos6.4, then failed on centos6.5. This was the magic trick that got me running again.

    It turns out upgrading to centos 6.5 removed the 32 bit gcc from my build machine.

    Thanks, Dinesh.

    ReplyDelete

Post a Comment

Popular posts from this blog

Parallel Operations With pl/pgSQL

Hi, I am pretty sure that, there will be a right heading for this post. For now, i am going with this. If you could suggest me proper heading, i will update it :-) OK. let me explain the situation. Then will let you know what i am trying to do here, and how i did it. Situation here is, We have a table, which we need to run update on “R” no.of records. The update query is using some joins to get the desired result, and do update the table.  To process these “R” no.of records, it is taking “H” no.of hours. That too, it’s giving load on the production server. So, we planned to run this UPDATE as batch process.  Per a batch process, we took “N” no.or records. To process this batch UPDATE, it is taking “S” no.of seconds. With the above batch process, production server is pretty stable, and doing great. So, we planned to run these Batch updates parallel.  I mean, “K” sessions, running different record UPDATEs. Of-course, we can also increase the Batch size here.  But

How To Send E-Mail From PostgreSQL

Hi , If you want to send E-Mails from PostgreSQL, then use the below Python 3.2 Script as below. I have used ActivePython 3.2 with PostgreSQL 9.1 for sending E-Mails from PostgreSQL. If you want to configure the Python 3.2 with PostgreSQL 9.1 then, please refer the below steps. http://manojadinesh.blogspot.in/2012/06/fatal-python-error-pyinitialize-unable.html Once, your Python 3.2 successful then follow the below steps to send an e-mail. Step 1 ===== postgres=# CREATE OR REPLACE FUNCTION public.send_email(_from Text,_password Text,smtp Text,port INT,receiver text, subject text, send_message text) RETURNS TEXT  LANGUAGE plpython3u AS $function$ import smtplib sender = _from receivers = receiver message = ("From: %s\nTo: %s\nSubject: %s\n\n %s"  % (_from,receiver,subject,send_message)) try:   smtpObj = smtplib.SMTP(smtp,port)   smtpObj.starttls()   smtpObj.login(_from, _password)   smtpObj.sendmail(sender, receivers,message)   print ('Successf

::Pipelined in Oracle as well in PostgreSQL::

Pipelined Table Functions:- [ORACLE] =========================== If you want to return multiple rows to the calling environment, then piplined table functions is prefred. It will increase the dbperformance as well. Ex:- Step 1: ----------- CREATE TABLE EMP(EMPNO INT,ENAME VARCHAR2(10),SAL INT); Step 2: ----------- Insert sample data. Step 3: ----------- Create an object for the row type casting. CREATE OR REPLACE TYPE emp_row AS OBJECT ( empno INT, ename VARCHAR2(20), SAL INT ); Step 4: ----------- Create a Return Type for the pipelined function. CREATE OR REPLACE TYPE emp_table_type AS TABLE OF emp_row; Step 5: ----------- CREATE OR REPLACE FUNCTION emp_pipe_function RETURN emp_table_type PIPELINED IS BEGIN FOR rec in (select * from emp) LOOP PIPE ROW (emp_row(rec.empno,rec.ename,rec.sal)); END LOOP; RETURN; END; Step 6: ---------- SQL> select * from table(emp_pipe_function); EMPNO ENAME SAL ---------- ----