Distcc

From ArchWiki
Jump to: navigation, search

Distcc is a program to distribute builds of C, C++, Objective C or Objective C++ code across several machines on a network to speed up building. It should always generate the same results as a local build, is simple to install and use, and is usually much faster than a local compile. Further, one can use it together with native Arch build tools such as makepkg.

Terms

master
The master is the computer which initiates the compilation.
slave
The slave accepts compilation requests send by the master. One can setup multiple slave systems or just a single one.


Getting started

Install the distcc package on all participating PCs in the distcc cluster. For other distros, or even operating systems including Windows through using Cygwin, refer to the distcc docs.

Configuration

Slaves

The configuration for the slave machine is stored in /etc/conf.d/distccd. At a minimum, configure the allowed address ranges in CIDR format:

DISTCC_ARGS="--allow 192.168.0.0/24"

A nice tool for converting address ranges to CIDR format can be found here: CIDR Utility Tool. Other commandline options can be defined as well. Refer to distcc(1).

Start distccd.service on every participating slave. To have distccd.service start at boot-up, enable it.

Master

For use with makepkg

Edit /etc/makepkg.conf in the following sections:

  1. The BUILDENV array will need to have distcc unbanged i.e. list it without exclamation point.
  2. Uncomment the DISTCC_HOSTS line and add the IP addresses of the slaves then a slash and the number of threads they are to use. The subsequent IP address/threads should be separated by a white space. This list is ordered from most powerful to least powerful (processing power).
  3. Adjust the MAKEFLAGS variable to correspond to the number of sum of the number of individual values specified for the max threads per server. In the example below, this is 5+3+3=11.
Note: The number of threads is commonly set as the number of cores plus 1. Do this on a per-server basis, not in the MAKEFLAGS variable.

Example using relevant lines:

BUILDENV=(distcc fakeroot color !ccache check !sign)
MAKEFLAGS="-j11"
DISTCC_HOSTS="192.168.0.2/5 192.168.0.3/3 192.168.0.4/3"
Note: The -march=native flag cannot be used in the CFLAGS and CXXFLAGS variables, otherwise distccd will not distribute work to other machines.

For use without makepkg

The minimal configuration for distcc on the master includes the setting of the available slaves. This can either be done by setting the addresses in the environment variable DISTCC_HOSTS or in either of the configuration files $DISTCC_HOSTS, $DISTCC_DIR/hosts, ~/.distcc/hosts or /etc/distcc/hosts.

Example for setting the slave address using DISTCC_HOSTS:

$ export DISTCC_HOSTS="192.168.0.3,lzo,cpp 192.168.0.4,lzo,cpp"
Note: This is a white space separated list.

Example for setting the slave addresses in the hosts configuration file:

~/.distcc/hosts
192.168.0.3,lzo,cpp 192.168.0.4,lzo,cpp

Instead of explicitly listing the server addresses one can also use the avahi zeroconf mode. To use this mode +zeroconf must be in place instead of the server addresses and the distcc daemons on the slaves have to be started using the --zeroconf option. Note that this option does not support the pump mode!

The examples add the following options to the address:

  • lzo: Enables LZO compression for this TCP or SSH host (slave).
  • cpp: Enables distcc-pump mode for this host (slave). Note: the build command must be wrapped in the pump script in order to start the include server.

A description for the pump mode can be found here: HOW DISTCC-PUMP MODE WORKS and distcc's pump mode: A New Design for Distributed C/C++ Compilation

To use distcc-pump mode for a slave, users must start the compilation using the pump script otherwise the compilation will fail.

Compile

With makepkg

Compile via makepkg as normal.

Without makepkg

To compile a source file using the distcc pump mode, use the following command:

$ pump distcc g++ -c hello_world.cpp

In this case the pump script will execute distcc which in turn calls g++ with "-c hello_world.cpp" as parameter.

To compile a Makefile project, first find out which variables are set by the compiler. For example in gzip-1.6, one can find the following line in the Makefile: CC = gcc -std=gnu99. Normally the variables are called CC for C projects and CXX for C++ projects. To compile the project using distcc it would look like this:

$ wget ftp://ftp.gnu.org/pub/gnu/gzip/gzip-1.6.tar.xz
$ tar xf gzip-1.6.tar.xz
$ cd gzip-1.6
$ ./configure
$ pump make -j2 CC="distcc gcc -std=gnu99"

This example would compile gzip using distcc's pump mode with two compile threads. For the correct -j setting have a look at What -j level to use?

Monitoring progress

Progress can be monitored via several methods.

  1. distccmon-text
  2. tailing log file

Invoke distccmon-text to check on compilation status:

$ distccmon-text
29291 Preprocess  probe_64.c                                 192.168.0.2[0]
30954 Compile     apic_noop.c                                192.168.0.2[0]
30932 Preprocess  kfifo.c                                    192.168.0.2[0]
30919 Compile     blk-core.c                                 192.168.0.2[1]
30969 Compile     i915_gem_debug.c                           192.168.0.2[3]
30444 Compile     block_dev.c                                192.168.0.3[1]
30904 Compile     compat.c                                   192.168.0.3[2]
30891 Compile     hugetlb.c                                  192.168.0.3[3]
30458 Compile     catalog.c                                  192.168.0.4[0]
30496 Compile     ulpqueue.c                                 192.168.0.4[2]
30506 Compile     alloc.c                                    192.168.0.4[0]

One can have this program run continuously by using watch or by appending a space followed by integer to the command which corresponds to the number of sec to wait for a repeat query:

$ watch distccmon-text

or

$ distccmon-text 2

One can also simply tail systemd journal log on daemon:

# journalctl -f -u distccd

"Cross Compiling" with distcc

32-bit x86 (i686)

There are currently two methods from which to select to have the ability of distcc distribution of tasks over a cluster building i686 packages from a native x86_64 environment. Neither is ideal, but to date, there are the only two methods documented on the wiki.

An ideal setup is one that uses the unmodified ARCH packages for distccd running only once one each node regardless of building from the native environment or from within a chroot AND one that works with makepkg. Again, this Utopian setup is not currently known.

A discussion thread has been started on the topic; feel free to contribute.

Chroot method (preferred)

Note: This method works, but is not very elegant requiring duplication of distccd on all nodes AND need to have a 32-bit chroots on all nodes.

Assuming the user has a 32-bit chroot setup and configured on each node of the distcc cluster, the strategy is to have two separate instances of distccd running on different ports on each node -- one runs in the native x86_64 environment and the other in the x86 chroot on a modified port. Start makepkg via a schroot command invoking makepkg.

Add port numbers to DISTCC_HOSTS on the i686 chroot

Append the port number defined eariler (3692) to each of the hosts in /opt/arch32/etc/makepkg.conf as follows:

DISTCC_HOSTS="192.168.1.101/5:3692 192.168.1.102/5:3692 192.168.1.103/3:3692"
Note: This only needs to be setup on the "master" i686 chroot. Where "master" is defined as the one from which the compilation will take place.
Invoke makepkg from the Native Environment

Setup schroot on the native x86_64 environment. Invoke makepkg to build an i686 package from the native x86_64 environment, simply by:

$ schroot -p -- makepkg -src

Multilib GCC method (not recommended)

See Makepkg#Build 32-bit packages on a 64-bit system.

Other architectures

Arch Linux ARM

When building on an Arch Linux ARM device, the developers highly recommend using the official project toolchains which should be installed on the x86_64 slave machine(s). Rather than manually managing these, the AUR provides all four toolchains as well as simple systemd service units:

Setup on the slave machine containing the toolchain is identical to #Slaves except that the name of the configuration file matches that of the respective package. For example, /etc/conf.d/distccd-armv7h.

A systemd service unit is provided for each respective package. For example, distccd-armv7h.service.

Additional toolchains

  • EmbToolkit: Tool for creating cross compilation tool chain; supports ARM and MIPS architectures; supports building of an LLVM based tool chain
  • crosstool-ng: Similar to EmbToolkit; supports more architectures (see website for more information)
  • Linaro: Provides tool chains for ARM development

The EmbToolkit provides a nice graphical configuration menu (make xconfig) for configuring the tool chain.

Troubleshooting

Journalctl

Use journalctl to find out what was going wrong:

$ journalctl $(which distccd) -e --since "5 min ago"

code 110

Make sure that the tool chain works for the user account under which the distcc daemon process gets started (default is nobody). The following will test if the tool chain works for user nobody. In /etc/passwd change the login for the nobody user to the following:

$ cat /etc/passwd
...
nobody:x:99:99:nobody:/:/bin/bash
...

Then cd into the directory containing the cross compiler binaries and try to execute the compiler:

# su nobody
$ ./gcc --version
bash: ./gcc: Permission denied

Users experiencing this error should make sure that groups permissions as described in #Other architectures are correctly in setup.

Make sure to change back /etc/passwd to its original state after these modifications.

Alternatively, use sudo without changing the shell in /etc/passwd.

 # sudo -u nobody gcc --version

Adjust log level

By default, distcc will log to /var/log/messages.log as it goes along. One trick (actually recommended in the distccd manpage) is to log to an alternative file directly. Again, one can locate this in RAM via /tmp. Another trick is to lower to log level of minimum severity of error that will be included in the log file. Useful if only wanting to see error messages rather than an entry for each connection. LEVEL can be any of the standard syslog levels, and in particular critical, error, warning, notice, info, or debug.

Either call distcc with the arguments mentioned here on the master or appended it to DISTCC_ARGS in /etc/conf.d/distccd on the slaves:

DISTCC_ARGS="--allow 192.168.0.0/24 --log-level error --log-file /tmp/distccd.log"

Limit HDD/SSD usage by relocating $HOME/.distcc

By default, distcc creates $HOME/.distcc which stores transient relevant info as it serves up work for nodes to compile. Create a directory named .distcc in RAM such as /tmp and soft link to it in $HOME. This will avoid needless HDD read/writes and is particularly important for SSDs.

$ mv $HOME/.distcc /tmp
$ ln -s /tmp/.distcc $HOME/.distcc

Use systemd to re-create this directory on a reboot (the soft link will remain until it is manually removed like any other file):

Create the following tmpfile.

/etc/tmpfiles.d/tmpfs-create.conf
d /tmp/.distcc 0755 <username> users -