launchpad-reviewers team mailing list archive

[Colin Watson <mp+452655@xxxxxxxxxxxxxxxxxx>]

Colin Watson has proposed merging ~cjwatson/launchpad:doc-services-build-farm into launchpad:master.

Commit message:
doc: Add high-level build farm documentation

Requested reviews:
  Launchpad code reviewers (launchpad-reviewers)

For more details, see:
https://code.launchpad.net/~cjwatson/launchpad/+git/launchpad/+merge/452655

Written collaboratively during a hackathon.
-- 
Your team Launchpad code reviewers is requested to review the proposed merge of ~cjwatson/launchpad:doc-services-build-farm into launchpad:master.

diff --git a/doc/reference/index.rst b/doc/reference/index.rst
index 0fd2897..836b6f2 100644
--- a/doc/reference/index.rst
+++ b/doc/reference/index.rst
@@ -15,6 +15,7 @@ Services
 .. toctree::
    :maxdepth: 1
 
+   services/build-farm
    services/signing
 
 Possibly out-of-date
diff --git a/doc/reference/services/build-farm.rst b/doc/reference/services/build-farm.rst
new file mode 100644
index 0000000..5dccb6b
--- /dev/null
+++ b/doc/reference/services/build-farm.rst
@@ -0,0 +1,164 @@
+Build farm
+==========
+
+Short description
+-----------------
+Builds and tests software for Ubuntu and related systems.
+
+Detailed description
+--------------------
+The build farm builds packages, such as debs, snaps, charms, and other
+formats for Ubuntu in a secure and isolated way.
+
+The build farm consists of a manager, a proxy for outbound communication,
+and various builder regions.
+Build jobs are created by other parts of Launchpad, such as a change to a
+git repository resulting in snap recipe builds.
+They are then scheduled onto builders as virtual machines by
+``buildd-manager``.
+
+``buildd-manager`` continuously monitors all builders in the farm, and when
+a builder is idle it chooses the next highest priority job for dispatch to
+the idle builder.
+It does this by using XML-RPC, passing relevant data to the builder, which
+in turn fetches all ingredients for executing the build.
+When a build has completed, ``buildd-manager`` gathers the resulting files
+and injects them into Launchpad via an upload queue processor, and then the
+builder is reset to a baseline state.
+
+The manner in which a build is executed is determined by the build type.
+This typically involves invoking some external tools; ``sbuild`` for Ubuntu
+package builds, ``snapcraft`` for snap builds, etc.
+
+Builders do not have direct access to the internet, but rather need to
+acquire an authentication token to be able to access a restricted set of
+URLs on the internet via a squid proxy.
+
+Builder regions are physically co-located and consist of machines of the
+same architecture family.
+Builder regions comprise per-architecture image builders, and each physical
+location contains a ``launchpad-vbuilder-manage`` instance which handles
+resets for all architectures within that location.
+
+Each builder region maintains clean VM images for its builders; these are
+built using ``glance-simplestreams-sync``, which automatically and
+periodically copies standard pre-built images from
+``cloud-images.ubuntu.com``, and ``launchpad-buildd-image-modifier``, which
+hooks into ``glance-simplestreams-sync`` to produce modified images with
+``launchpad-buildd`` installed.
+
+Documentation
+-------------
+* `Documentation for launchpad-buildd <https://launchpad-buildd.readthedocs.io/en/latest/index.html>`_
+* `Documentation for the Launchpad CI runner <https://lpci.readthedocs.io/en/latest/>`_
+* `Documentation for Launchpad CI <https://help.launchpad.net/Code/ContinuousIntegration>`_
+
+Git repositories
+----------------
+* `buildd-manager <https://git.launchpad.net/launchpad/tree/lib/lp/buildmaster/>`_
+* `launchpad-buildd <https://git.launchpad.net/launchpad-buildd>`_
+* `Launchpad CI runner <https://git.launchpad.net/lpci>`_
+* `glance-simplestreams-sync charm <https://git.launchpad.net/~launchpad/charm-glance-simplestreams-sync/tree/?h=scalingstack>`_
+* `launchpad-buildd-image-modifier charm <https://git.launchpad.net/charm-launchpad-buildd-image-modifier>`_
+* `launchpad-vbuilder-manage <https://git.launchpad.net/launchpad-vbuilder-manage>`_
+
+Bug trackers
+------------
+* https://bugs.launchpad.net/launchpad-project/+bugs?field.tag=soyuz-build
+* https://bugs.launchpad.net/launchpad-buildd
+* https://bugs.launchpad.net/lpci
+* https://bugs.launchpad.net/charm-glance-simplestreams-sync
+* https://bugs.launchpad.net/charm-launchpad-buildd-image-modifier
+* https://bugs.launchpad.net/launchpad-vbuilder-manage
+
+Deployment
+----------
+* `Deployment of launchpad-buildd <https://launchpad-buildd.readthedocs.io/en/latest/how-to/deployment.html>`_
+* `Production deployment notes for launchpad-buildd <https://launchpad-buildd.readthedocs.io/en/latest/explanation/deployment.html>`_
+* `vbuilder Mojo spec <https://git.launchpad.net/~launchpad/launchpad-mojo-specs/+git/private/tree/vbuilder?h=vbuilder>`_
+* `lp-builder-proxy Mojo spec <https://git.launchpad.net/launchpad-mojo-specs/tree/lp-builder-proxy/>`_
+
+Related specifications
+----------------------
+`LP113: Threat model for builds <https://docs.google.com/document/d/1im8CMxLRNxtt5H0zv461kSYSflN-YlxJ1UZG8_53D9A>`_
+
+Log files
+---------
+See https://wiki.canonical.com/Launchpad/FreshLogs.
+
+Production
+~~~~~~~~~~
+
+buildd-manager
+^^^^^^^^^^^^^^
+
+* ``rless buildd-manager.lp.internal::lp-logs/buildd-manager.log``
+* ``rless buildd-manager.lp.internal::lp-logs/process-build-uploads.log``
+* ``rless buildd-manager.lp.internal::lp-logs/buildd-retry-depwait.log``
+
+vbuilder-manage
+^^^^^^^^^^^^^^^
+
+Builder reset logs.
+Each celery worker has a different log file, named ``celery.ppareset-*.log``.
+
+* ``rsync -v vbuilder-manage-lcy02.lp.internal::vbuilder-manage-logs/``
+* ``rsync -v vbuilder-manage-bos01.lp.internal::vbuilder-manage-logs/``
+* ``rsync -v vbuilder-manage-bos02.lp.internal::vbuilder-manage-logs/``
+* ``rsync -v vbuilder-manage-bos03.lp.internal::vbuilder-manage-logs/``
+
+builder-proxy (auth)
+^^^^^^^^^^^^^^^^^^^^
+
+* ``rless builder-proxy-auth.lp.internal::rutabaga-logs/rutabaga-access.log``
+* ``rless builder-proxy-auth.lp.internal::rutabaga-logs/rutabaga-error.log``
+* ``rless builder-proxy-auth.lp.internal::rutabaga-logs/rutabaga-purge.log``
+
+builder-proxy (squid) 
+^^^^^^^^^^^^^^^^^^^^^
+
+* ``rless 10.131.48.38::squid-logs/access.log``
+* ``rless 10.131.48.38::squid-logs/cache.log``
+* ``rless 10.131.48.24::squid-logs/access.log``
+* ``rless 10.131.48.24::squid-logs/cache.log``
+
+Staging
+~~~~~~~
+
+buildd-manager
+^^^^^^^^^^^^^^
+
+* ``rless 10.132.54.143::lp-logs/buildd-manager.log``
+* ``rless 10.132.54.143::lp-logs/process-build-uploads.log``
+* ``rless 10.132.54.143::lp-logs/buildd-retry-depwait.log``
+
+builder-proxy (auth)
+^^^^^^^^^^^^^^^^^^^^
+
+* ``rless builder-proxy-auth.staging.lp.internal::rutabaga-logs/rutabaga-access.log``
+* ``rless builder-proxy-auth.staging.lp.internal::rutabaga-logs/rutabaga-error.log``
+* ``rless builder-proxy-auth.staging.lp.internal::rutabaga-logs/rutabaga-purge.log``
+
+builder-proxy (squid) 
+^^^^^^^^^^^^^^^^^^^^^
+
+* ``rless 10.132.224.179::squid-logs/access.log``
+* ``rless 10.132.224.179::squid-logs/cache.log``
+* ``rless 10.132.224.16::squid-logs/access.log``
+* ``rless 10.132.224.16::squid-logs/cache.log``
+
+Monitoring
+----------
+The "Build farm" section of the `Launchpad dash <https://grafana.admin.canonical.com/d/oIhMaXhMk/launchpad-dash>`_.
+
+Search for "build" in https://git.launchpad.net/canonical-is-prometheus/tree/ols/launchpad.rules.
+
+Common support cases
+--------------------
+See `Launchpad's playbook for support rotation <https://wiki.canonical.com/Launchpad/SupportRotation#Builder>`_
+
+More information
+----------------
+
+* `Live overview of Launchpad's build farm <https://launchpad.net/builders>`_
+* `Launchpad services diagram <https://viewer.diagrams.net/?tags=%7B%7D&highlight=0000ff&edit=_blank&layers=1&nav=1&page-id=14glVH8XSJX-2FxTRWny#G1j-yk3c4mzYfMC79Y-uo9__u93pLWkiMi>`_