Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add documentation for chaining Spack instances #11152

Merged
merged 5 commits into from
Apr 10, 2019
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
98 changes: 98 additions & 0 deletions lib/spack/docs/chain.rst
Original file line number Diff line number Diff line change
@@ -0,0 +1,98 @@
.. Copyright 2013-2019 Lawrence Livermore National Security, LLC and other
Spack Project Developers. See the top-level COPYRIGHT file for details.

SPDX-License-Identifier: (Apache-2.0 OR MIT)

.. chain:

============================
Chaining Spack Installations
============================

You can point your Spack installation to another installation to use any
packages that are installed there. To register the other Spack instance,
you can add it as an entry to ``config.yaml``:

.. code-block:: yaml

config:
upstreams:
spack-instance-1:
install_tree: /path/to/other/spack/opt/spack
spack-instance-2:
install_tree: /path/to/another/spack/opt/spack

``install_tree`` must point to the ``opt/spack`` directory inside of the
Spack base directory.

Once the upstream Spack instance has been added, ``spack find`` will
automatically check the upstream instance when querying installed packages,
and new package installations for the local Spack install will use any
dependencies that are installed in the upstream instance.

This other instance of Spack has no knowledge of the local Spack instance
and may not have the same permissions or ownership as the local Spack instance.
This has the following consequences:

#. Upstream Spack instances are not locked. Therefore it is up to users to
make sure that the local instance is not using an upstream instance when it
is being modified.

#. Users should not uninstall packages from the upstream instance. Since the
upstream instance doesn't know about the local instance, it cannot prevent
the uninstallation of packages which the local instance depends on.

Other details about upstream installations:

#. If a package is installed both locally and upstream, the local installation
will always be used as a dependency. This can occur if the local Spack
installs a package which is not present in the upstream, but later on the
upstream Spack instance also installs that package.

#. If an upstream Spack instance registers and installs an external package,
the local Spack instance will treat this the same as a Spack-installed
package. This feature will only work if the upstream Spack instance
includes the upstream functionality (i.e. if its commit is after March
27, 2019).

---------------------------------------
Using Multiple Upstream Spack Instances
---------------------------------------

A single Spack instance can use multiple upstream Spack installations. Spack
will search upstream instances in the order you list them in your
configuration. If your installation refers to instances X and Y, in that order,
then instance X must list Y as an upstream in its own ``config.yaml``.

-----------------------------------
Using Modules for Upstream Packages
-----------------------------------

The local Spack instance does not generate modules for packages which are
installed upstream. The local Spack instance can be configured to use the
modules generated by the upstream Spack instance.

There are two requirements to use the modules created by an upstream Spack
instance: firstly the upstream instance must do a ``spack module tcl refresh``,
which generates an index file that maps installed packages to their modules;
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is probably unrelated to this PR but would it not make more sense to generate/update the index whenever a module is generated instead of forcing users to refresh them?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It would. I suppose I originally didn't want to do the extra work in a Spack instance if it wasn't being used as an upstream, but it's not a lot of work.

secondly, the local Spack instance must add a ``modules`` entry to the
configuration:

.. code-block:: yaml

config:
upstreams:
spack-instance-1:
install_tree: /path/to/other/spack/opt/spack
modules:
tcl: /path/to/other/spack/share/spack/modules

Each time new packages are installed in the upstream Spack instance, the
upstream Spack maintainer should run ``spack module tcl refresh`` (or the
corresponding command for the type of module they intend to use).

.. note::

Spack can generate modules that :ref:`automatically load
<autoloading-dependencies>` the modules of dependency packages. Spack cannot
currently do this for modules in upstream packages.
1 change: 1 addition & 0 deletions lib/spack/docs/index.rst
Original file line number Diff line number Diff line change
Expand Up @@ -71,6 +71,7 @@ or refer to the full manual below.
binary_caches
command_index
package_list
chain

.. toctree::
:maxdepth: 2
Expand Down
3 changes: 3 additions & 0 deletions lib/spack/docs/module_file_support.rst
Original file line number Diff line number Diff line change
Expand Up @@ -600,6 +600,9 @@ The configuration above will generate dotkit module files that will not contain
modifications to either ``CPATH`` or ``LIBRARY_PATH`` and environment module
files that instead will contain these modifications.


.. _autoloading-dependencies:

"""""""""""""""""""""
Autoload dependencies
"""""""""""""""""""""
Expand Down