| PostgreSQL 9.1.3 Documentation | ||||
|---|---|---|---|---|
| Prev | Up | Chapter 15. Installation from Source Code | Next | |
This section documents additional platform-specific issues regarding the installation and setup of PostgreSQL. Be sure to read the installation instructions, and in particular Section 15.2 as well. Also, check Chapter 30 regarding the interpretation of regression test results.
Platforms that are not covered here have no known platform-specific installation issues.
PostgreSQL works on AIX, but getting it installed properly can be challenging. AIX versions from 4.3.3 to 6.1 are considered supported. You can use GCC or the native IBM compiler xlc. In general, using recent versions of AIX and PostgreSQL helps. Check the build farm for up to date information about which versions of AIX are known to work.
The minimum recommended fix levels for supported AIX versions are:
Maintenance Level 11 + post ML11 bundle
Maintenance Level 9 + post ML9 bundle
Technology Level 10 Service Pack 3
Technology Level 7
Base Level
To check your current fix level, use oslevel -r in AIX 4.3.3 to AIX 5.2 ML 7, or oslevel -s in later versions.
Use the following configure flags in addition to your own if you have installed Readline or libz in /usr/local: --with-includes=/usr/local/include --with-libraries=/usr/local/lib.
On AIX 5.3, there have been some problems getting PostgreSQL to compile and run using GCC.
You will want to use a version of GCC subsequent to 3.3.2, particularly if you use a prepackaged version. We had good success with 4.0.1. Problems with earlier versions seem to have more to do with the way IBM packaged GCC than with actual issues with GCC, so that if you compile GCC yourself, you might well have success with an earlier version of GCC.
AIX 5.3 has a problem where sockaddr_storage is not defined to be large enough. In version 5.3, IBM increased the size of sockaddr_un, the address structure for Unix-domain sockets, but did not correspondingly increase the size of sockaddr_storage. The result of this is that attempts to use Unix-domain sockets with PostgreSQL lead to libpq overflowing the data structure. TCP/IP connections work OK, but not Unix-domain sockets, which prevents the regression tests from working.
The problem was reported to IBM, and is recorded as bug report PMR29657. If you upgrade to maintenance level 5300-03 or later, that will include this fix. A quick workaround is to alter _SS_MAXSIZE to 1025 in /usr/include/sys/socket.h. In either case, recompile PostgreSQL once you have the corrected header file.
PostgreSQL relies on the system's getaddrinfo function
to parse IP addresses in listen_addresses,
pg_hba.conf, etc. Older versions of AIX have assorted
bugs in this function. If you have problems related to these settings,
updating to the appropriate AIX fix level shown above
should take care of it.
One user reports:
When implementing PostgreSQL version 8.1 on AIX 5.3, we periodically ran into problems where the statistics collector would "mysteriously" not come up successfully. This appears to be the result of unexpected behavior in the IPv6 implementation. It looks like PostgreSQL and IPv6 do not play very well together on AIX 5.3.
Any of the following actions "fix" the problem.
Delete the IPv6 address for localhost:
(as root) # ifconfig lo0 inet6 ::1/0 delete
Remove IPv6 from net services. The file /etc/netsvc.conf on AIX is roughly equivalent to /etc/nsswitch.conf on Solaris/Linux. The default, on AIX, is thus:
hosts=local,bind
Replace this with:
hosts=local4,bind4
to deactivate searching for IPv6 addresses.
| Warning |
This is really a workaround for problems relating to immaturity of IPv6 support, which improved visibly during the course of AIX 5.3 releases. It has worked with AIX version 5.3, but does not represent an elegant solution to the problem. It has been reported that this workaround is not only unnecessary, but causes problems on AIX 6.1, where IPv6 support has become more mature. |
AIX can be somewhat peculiar with regards to the way it does memory management. You can have a server with many multiples of gigabytes of RAM free, but still get out of memory or address space errors when running applications. One example is createlang failing with unusual errors. For example, running as the owner of the PostgreSQL installation:
-bash-3.00$ createlang plperl template1 createlang: language installation failed: ERROR: could not load library "/opt/dbs/pgsql748/lib/plperl.so": A memory address is not in the address space for the process.
Running as a non-owner in the group possessing the PostgreSQL installation:
-bash-3.00$ createlang plperl template1 createlang: language installation failed: ERROR: could not load library "/opt/dbs/pgsql748/lib/plperl.so": Bad address
Another example is out of memory errors in the PostgreSQL server logs, with every memory allocation near or greater than 256 MB failing.
The overall cause of all these problems is the default bittedness and memory model used by the server process. By default, all binaries built on AIX are 32-bit. This does not depend upon hardware type or kernel in use. These 32-bit processes are limited to 4 GB of memory laid out in 256 MB segments using one of a few models. The default allows for less than 256 MB in the heap as it shares a single segment with the stack.
In the case of the createlang example, above, check your umask and the permissions of the binaries in your PostgreSQL installation. The binaries involved in that example were 32-bit and installed as mode 750 instead of 755. Due to the permissions being set in this fashion, only the owner or a member of the possessing group can load the library. Since it isn't world-readable, the loader places the object into the process' heap instead of the shared library segments where it would otherwise be placed.
The "ideal" solution for this is to use a 64-bit build of PostgreSQL, but that is not always practical, because systems with 32-bit processors can build, but not run, 64-bit binaries.
If a 32-bit binary is desired, set LDR_CNTRL to MAXDATA=0xn0000000, where 1 <= n <= 8, before starting the PostgreSQL server, and try different values and postgresql.conf settings to find a configuration that works satisfactorily. This use of LDR_CNTRL tells AIX that you want the server to have MAXDATA bytes set aside for the heap, allocated in 256 MB segments. When you find a workable configuration, ldedit can be used to modify the binaries so that they default to using the desired heap size. PostgreSQL can also be rebuilt, passing configure LDFLAGS="-Wl,-bmaxdata:0xn0000000" to achieve the same effect.
For a 64-bit build, set OBJECT_MODE to 64 and pass CC="gcc -maix64" and LDFLAGS="-Wl,-bbigtoc" to configure. (Options for xlc might differ.) If you omit the export of OBJECT_MODE, your build may fail with linker errors. When OBJECT_MODE is set, it tells AIX's build utilities such as ar, as, and ld what type of objects to default to handling.
By default, overcommit of paging space can happen. While we have not seen this occur, AIX will kill processes when it runs out of memory and the overcommit is accessed. The closest to this that we have seen is fork failing because the system decided that there was not enough memory for another process. Like many other parts of AIX, the paging space allocation method and out-of-memory kill is configurable on a system- or process-wide basis if this becomes a problem.
"Large Program Support", AIX Documentation: General Programming Concepts: Writing and Debugging Programs.
"Program Address Space Overview", AIX Documentation: General Programming Concepts: Writing and Debugging Programs.
"Performance Overview of the Virtual Memory Manager (VMM)", AIX Documentation: Performance Management Guide.
"Page Space Allocation", AIX Documentation: Performance Management Guide.
"Paging-space thresholds tuning", AIX Documentation: Performance Management Guide.
Developing and Porting C and C++ Applications on AIX, IBM Redbook.
PostgreSQL can be built using Cygwin, a Linux-like environment for Windows, but that method is inferior to the native Windows build (see Chapter 16) and running a server under Cygwin is no longer recommended.
When building from source, proceed according to the normal installation procedure (i.e., ./configure; make; etc.), noting the following-Cygwin specific differences:
Set your path to use the Cygwin bin directory before the Windows utilities. This will help prevent problems with compilation.
The GNU make command is called make, not gmake.
The adduser command is not supported; use the appropriate user management application on Windows NT, 2000, or XP. Otherwise, skip this step.
The su command is not supported; use ssh to simulate su on Windows NT, 2000, or XP. Otherwise, skip this step.
OpenSSL is not supported.
Start cygserver for shared memory support. To do this, enter the command /usr/sbin/cygserver &. This program needs to be running anytime you start the PostgreSQL server or initialize a database cluster (initdb). The default cygserver configuration may need to be changed (e.g., increase SEMMNS) to prevent PostgreSQL from failing due to a lack of system resources.
Building might fail on some systems where a locale other than C is in use. To fix this, set the locale to C by doing export LANG=C.utf8 before building, and then setting it back to the previous setting, after you have installed PostgreSQL.
The parallel regression tests (make check)
can generate spurious regression test failures due to
overflowing the listen() backlog queue
which causes connection refused errors or hangs. You can limit
the number of connections using the make
variable MAX_CONNECTIONS thus:
make MAX_CONNECTIONS=5 check
(On some systems you can have up to about 10 simultaneous connections).
It is possible to install cygserver and the PostgreSQL server as Windows NT services. For information on how to do this, please refer to the README document included with the PostgreSQL binary package on Cygwin. It is installed in the directory /usr/share/doc/Cygwin.
PostgreSQL 7.3+ should work on Series 700/800 PA-RISC machines running HP-UX 10.X or 11.X, given appropriate system patch levels and build tools. At least one developer routinely tests on HP-UX 10.20, and we have reports of successful installations on HP-UX 11.00 and 11.11.
Aside from the PostgreSQL source distribution, you will need GNU make (HP's make will not do), and either GCC or HP's full ANSI C compiler. If you intend to build from Git sources rather than a distribution tarball, you will also need Flex (GNU lex) and Bison (GNU yacc). We also recommend making sure you are fairly up-to-date on HP patches. At a minimum, if you are building 64 bit binaries on on HP-UX 11.11 you may need PHSS_30966 (11.11) or a successor patch otherwise initdb may hang:
PHSS_30966 s700_800 ld(1) and linker tools cumulative patch
On general principles you should be current on libc and ld/dld patches, as well as compiler patches if you are using HP's C compiler. See HP's support sites such as http://itrc.hp.com and ftp://us-ffs.external.hp.com/ for free copies of their latest patches.If you are building on a PA-RISC 2.0 machine and want to have 64-bit binaries using GCC, you must use GCC 64-bit version. GCC binaries for HP-UX PA-RISC and Itanium are available from http://www.hp.com/go/gcc. Don't forget to get and install binutils at the same time.
If you are building on a PA-RISC 2.0 machine and want the compiled binaries to run on PA-RISC 1.1 machines you will need to specify +DAportable in CFLAGS.
If you are building on a HP-UX Itanium machine, you will need the latest HP ANSI C compiler with its dependent patch or successor patches:
PHSS_30848 s700_800 HP C Compiler (A.05.57)
PHSS_30849 s700_800 u2comp/be/plugin library Patch
If you have both HP's C compiler and GCC's, then you might want to explicitly select the compiler to use when you run configure:
./configure CC=cc
for HP's C compiler, or
./configure CC=gcc
for GCC. If you omit this setting, then configure will pick gcc if it has a choice.
The default install target location is /usr/local/pgsql, which you might want to change to something under /opt. If so, use the --prefix switch to configure.
In the regression tests, there might be some low-order-digit differences in the geometry tests, which vary depending on which compiler and math library versions you use. Any other error is cause for suspicion.
PostgreSQL has been reported to run successfully on MIPS r8000, r10000 (both ip25 and ip27) and r12000(ip35) processors, running IRIX 6.5.5m, 6.5.12, 6.5.13, and 6.5.26 with MIPSPro compilers version 7.30, 7.3.1.2m, 7.3, and 7.4.4m.
You will need the MIPSPro full ANSI C compiler. There are
problems trying to build with GCC. It is a known GCC bug (not
fixed as of version 3.0) related to using functions that return
certain kinds of structures. This bug affects functions like
inet_ntoa, inet_lnaof, inet_netof, inet_makeaddr,
and semctl. It is supposed to be fixed by forcing
code to link those functions with libgcc, but this has not been
tested yet.
It is known that version 7.4.1m of the MIPSPro compiler generates incorrect code. The symptom is "invalid primary checkpoint record" when trying to start the database.) Version 7.4.4m is OK; the status of intermediate versions is uncertain.
There may be a compilation problem like the following:
cc-1020 cc: ERROR File = pqcomm.c, Line = 427
The identifier "TCP_NODELAY" is undefined.
if (setsockopt(port->sock, IPPROTO_TCP, TCP_NODELAY,Some versions include TCP definitions in sys/xti.h, so it is necessary to add #include <sys/xti.h> in src/backend/libpq/pqcomm.c and in src/interfaces/libpq/fe-connect.c. If you encounter this, please let us know so we can develop a proper fix.
In the regression tests, there might be some low-order-digit differences in the geometry tests, depending on which FPU are you using. Any other error is cause for suspicion.
PostgreSQL for Windows can be built using MinGW, a Unix-like build environment for Microsoft operating systems, or using Microsoft's Visual C++ compiler suite. The MinGW build variant uses the normal build system described in this chapter; the Visual C++ build works completely differently and is described in Chapter 16. It is a fully native build and uses no additional software like MinGW. A ready-made installer is available on the main PostgreSQL web site.
The native Windows port requires a 32 or 64-bit version of Windows 2000 or later. Earlier operating systems do not have sufficient infrastructure (but Cygwin may be used on those). MinGW, the Unix-like build tools, and MSYS, a collection of Unix tools required to run shell scripts like configure, can be downloaded from http://www.mingw.org/. Neither is required to run the resulting binaries; they are needed only for creating the binaries.
To build 64 bit binaries using MinGW, install the 64 bit tool set from http://mingw-w64.sourceforge.net/, put its bin directory in the PATH, and run configure with the --host=x86_64-w64-mingw option.
After you have everything installed, it is suggested that you run psql under CMD.EXE, as the MSYS console has buffering issues.
If PostgreSQL on Windows crashes, it has the ability to generate minidumps that can be used to track down the cause for the crash, similar to core dumps on Unix. These dumps can be read using the Windows Debugger Tools or using Visual Studio. To enable the generation of dumps on Windows, create a subdirectory named crashdumps inside the cluster data directory. The dumps will then be written into this directory with a unique name based on the identifier of the crashing process and the current time of the crash.
PostgreSQL can be built on SCO UnixWare 7 and SCO OpenServer 5. On OpenServer, you can use either the OpenServer Development Kit or the Universal Development Kit. However, some tweaking may be needed, as described below.
You should locate your copy of the SCO Skunkware CD. The Skunkware CD is included with UnixWare 7 and current versions of OpenServer 5. Skunkware includes ready-to-install versions of many popular programs that are available on the Internet. For example, gzip, gunzip, GNU Make, Flex, and Bison are all included. For UnixWare 7.1, this CD is now labeled "Open License Software Supplement". If you do not have this CD, the software on it is available from http://www.sco.com/skunkware/.
Skunkware has different versions for UnixWare and OpenServer. Make sure you install the correct version for your operating system, except as noted below.
On UnixWare 7.1.3 and beyond, the GCC compiler is included on the UDK CD as is GNU Make.
You need to use the GNU Make program, which is on the Skunkware CD. By default, it installs as /usr/local/bin/make. To avoid confusion with the SCO make program, you may want to rename GNU make to gmake.
As of UnixWare 7.1.3 and above, the GNU Make program is is the OSTK portion of the UDK CD, and is in /usr/gnu/bin/gmake.
The Readline library is on the Skunkware CD. But it is not included on the UnixWare 7.1 Skunkware CD. If you have the UnixWare 7.0.0 or 7.0.1 Skunkware CDs, you can install it from there. Otherwise, try http://www.sco.com/skunkware/.
By default, Readline installs into /usr/local/lib and /usr/local/include. However, the PostgreSQL configure program will not find it there without help. If you installed Readline, then use the following options to configure:
./configure --with-libraries=/usr/local/lib --with-includes=/usr/local/include
If you are using the new Universal Development Kit (UDK) compiler on OpenServer, you need to specify the locations of the UDK libraries:
./configure --with-libraries=/udk/usr/lib --with-includes=/udk/usr/include
Putting these together with the Readline options from above:
./configure --with-libraries="/udk/usr/lib /usr/local/lib" --with-includes="/udk/usr/include /usr/local/include"
By default, the PostgreSQL man pages are installed into /usr/local/pgsql/man. By default, UnixWare does not look there for man pages. To be able to read them you need to modify the MANPATH variable in /etc/default/man, for example:
MANPATH=/usr/lib/scohelp/%L/man:/usr/dt/man:/usr/man:/usr/share/man:scohelp:/usr/local/man:/usr/local/pgsql/man
On OpenServer, some extra research needs to be invested to make the man pages usable, because the man system is a bit different from other platforms. Currently, PostgreSQL will not install them at all.
For compilers earlier than the one released with OpenUNIX 8.0.0 (UnixWare 7.1.2), including the 7.1.1b Feature Supplement, you may need to specify -Xb in CFLAGS or the CC environment variable. The indication of this is an error in compiling tuplesort.c referencing inline functions. Apparently there was a change in the 7.1.2(8.0.0) compiler and beyond.
For threading, youmust use -Kpthread
on all libpq-using programs. libpq
uses pthread_* calls, which are only
available with the
-Kpthread/-Kthread flag.
PostgreSQL is well-supported on Solaris. The more up to date your operating system, the fewer issues you will experience; details below.
Note that PostgreSQL is bundled with Solaris 10 (from update 2). Official packages are also available on http://pgfoundry.org/projects/solarispackages/. Packages for older Solaris versions (8, 9) you can be obtained from http://www.sunfreeware.com/ or http://www.blastwave.org/.
You can build with either GCC or Sun's compiler suite. For better code optimization, Sun's compiler is strongly recommended on the SPARC architecture. We have heard reports of problems when using GCC 2.95.1; GCC 2.95.3 or later is recommended. If you are using Sun's compiler, be careful not to select /usr/ucb/cc; use /opt/SUNWspro/bin/cc.
You can download Sun Studio from http://developers.sun.com/sunstudio/downloads/. Many of GNU tools are integrated into Solaris 10, or they are present on the Solaris companion CD. If you like packages for older version of Solaris, you can find these tools at http://www.sunfreeware.com or http://www.blastwave.org. If you prefer sources, look at http://www.gnu.org/order/ftp.html.
When you build PostgreSQL with OpenSSL support you might get compilation errors in the following files:
src/backend/libpq/crypt.c
src/backend/libpq/password.c
src/interfaces/libpq/fe-auth.c
src/interfaces/libpq/fe-connect.c
This is because of a namespace conflict between the standard /usr/include/crypt.h header and the header files provided by OpenSSL.
Upgrading your OpenSSL installation to version 0.9.6a fixes this problem. Solaris 9 and above has a newer version of OpenSSL.
If configure complains about a failed test program, this is probably a case of the run-time linker being unable to find some library, probably libz, libreadline or some other non-standard library such as libssl. To point it to the right location, set the LDFLAGS environment variable on the configure command line, e.g.,
configure ... LDFLAGS="-R /usr/sfw/lib:/opt/sfw/lib:/usr/local/lib"
See the ld man page for more information.
On Solaris 7 and older, the 64-bit version of libc has a buggy
vsnprintf routine, which leads to erratic
core dumps in PostgreSQL. The simplest known workaround is to
force PostgreSQL to use its own version of vsnprintf rather than
the library copy. To do this, after you
run configure edit a file produced by
configure:
In src/Makefile.global, change the line
LIBOBJS =
to read
LIBOBJS = snprintf.o
(There might be other files already listed in this variable. Order does not matter.) Then build as usual.
On the SPARC architecture, Sun Studio is strongly recommended for compilation. Try using the -xO5 optimization flag to generate significantly faster binaries. Do not use any flags that modify behavior of floating-point operations and errno processing (e.g., -fast). These flags could raise some nonstandard PostgreSQL behavior for example in the date/time computing.
If you do not have a reason to use 64-bit binaries on SPARC, prefer the 32-bit version. The 64-bit operations are slower and 64-bit binaries are slower than the 32-bit variants. And on other hand, 32-bit code on the AMD64 CPU family is not native, and that is why 32-bit code is significant slower on this CPU family.
Some tricks for tuning PostgreSQL and Solaris for performance can be found at http://www.sun.com/servers/coolthreads/tnb/applications_postgresql.jsp. This article is primary focused on T2000 platform, but many of the recommendations are also useful on other hardware with Solaris.
Yes, using DTrace is possible. See Section 27.4 for further information. You can also find more information in this article: http://blogs.sun.com/robertlor/entry/user_level_dtrace_probes_in.
If you see the linking of the postgres executable abort with an error message like:
Undefined first referenced symbol in file AbortTransaction utils/probes.o CommitTransaction utils/probes.o ld: fatal: Symbol referencing errors. No output written to postgres collect2: ld returned 1 exit status gmake: *** [postgres] Error 1
your DTrace installation is too old to handle probes in static functions. You need Solaris 10u4 or newer.