MaverickCDBuildSpeed

Differences between revisions 1 and 4 (spanning 3 versions)
Revision 1 as of 2010-05-03 14:35:09
Size: 2603
Editor: eth0
Comment:
Revision 4 as of 2010-05-21 09:45:15
Size: 2943
Editor: 82-69-40-219
Comment: draft
Deletions are marked like this. Additions are marked like this.
Line 5: Line 5:
 * '''Contributors''':
 * '''Packages affected''':
 * '''Contributors''': LaMontJones, CodySomerville, ColinWatson, ...
 * '''Packages affected''': `livecd-rootfs`
Line 10: Line 10:
This should provide an overview of the issue/functionality/change proposed here. Focus here on what will actually be DONE, summarising that so that other people don't have to read the whole spec. See also CategorySpec for examples. CD build speed has been creeping back up of late; attack it again.
Line 14: Line 14:
This section should include a paragraph describing the end-user impact of this change. It is meant to be included in the release notes of the first release in which it is implemented. (Not all of these will actually be included in the release notes, at the release manager's discretion; but writing them is a useful exercise.)

It is mandatory.
Not a user-visible feature, so no release note required.
Line 20: Line 18:
This should cover the _why_: why is this change being proposed, what justifies it, where we see this justified.

== User stories ==

== Assumptions ==
CD build times are a bottleneck in the release process. We have [[UbuntuSpec:cd-build-process|worked on this before]], but we build many more images now and the problem has once again become acute. At the moment, live filesystem build times seem to be the worst offenders.
Line 28: Line 22:
You can have subsections that better describe specific parts of the issue. === Primary work ===
Line 30: Line 24:
== Implementation == To permit future measurements, we will change `livecd-rootfs` to log the total time taken for live filesystem builds, as well as any other intermediate timings which may be useful.
Line 32: Line 26:
This section should describe a plan of action (the "how") to implement the changes discussed. Could include subsections like: Building on a tmpfs speeds up live filesystem builds substantially (down to 7 minutes on amd64, down from 21 minutes to 16 minutes on i386). This is only practical for non-DVD builds, but we will do it for other builds.
Line 34: Line 28:
=== UI Changes === Early stages of live filesystem builds are common to multiple flavours, particularly the base system `debootstrap` and the desktop element of DVD builds. We will profile the results of rsyncing this to a safe location along with an archive timestamp, and simply copying it back rather than rebuilding it if the archive timestamp hasn't changed. If profiling suggests that this is beneficial then we will do it. (Cody Somerville says that saving the results of `debootstrap` is a net gain for `live-helper`.)
Line 36: Line 30:
Should cover changes required to the UI, or specific UI that is required to implement this Cody Somerville says that some degree of parallelisation speeds up live filesystem builds for OEM Services. We will investigate whether this is also true for `livecd-rootfs`, and if so what maximum degree of parallelisation we can allow.
Line 38: Line 32:
=== Code Changes === We will add a facility to remove the `/usr/share/doc` report and the `fdupes` run from release builds.
Line 40: Line 34:
Code changes should include an overview of what needs to change, and in some cases even the specific details. We will add a dpkg option to turn off I/O safety (fsync et al) and use that during live filesystem builds.
Line 42: Line 36:
=== Migration === === Secondary work ===
Line 44: Line 38:
Include:
 * data migration, if any
 * redirects from old URLs to new ones, if any
 * how users will be pointed to the new way of doing things, if necessary.
We will evaluate `live-helper` as a replacement for `livecd-rootfs`. In order to commit to this, we would at minimum need to do a file-by-file comparison of the output and sign off on any differences; a moderate amount of customisation work will almost certainly be required.
Line 55: Line 46:
== Unresolved issues ==

This should highlight any issues that should be addressed in further specifications, and not problems with the specification itself; since any specification with problems cannot be approved.

== BoF agenda and discussion ==

Use this section to take notes during the BoF; if you keep it in the approved spec, use it for summarising what was discussed and note any options that were rejected.

Summary

CD build speed has been creeping back up of late; attack it again.

Release Note

Not a user-visible feature, so no release note required.

Rationale

CD build times are a bottleneck in the release process. We have worked on this before, but we build many more images now and the problem has once again become acute. At the moment, live filesystem build times seem to be the worst offenders.

Design

Primary work

To permit future measurements, we will change livecd-rootfs to log the total time taken for live filesystem builds, as well as any other intermediate timings which may be useful.

Building on a tmpfs speeds up live filesystem builds substantially (down to 7 minutes on amd64, down from 21 minutes to 16 minutes on i386). This is only practical for non-DVD builds, but we will do it for other builds.

Early stages of live filesystem builds are common to multiple flavours, particularly the base system debootstrap and the desktop element of DVD builds. We will profile the results of rsyncing this to a safe location along with an archive timestamp, and simply copying it back rather than rebuilding it if the archive timestamp hasn't changed. If profiling suggests that this is beneficial then we will do it. (Cody Somerville says that saving the results of debootstrap is a net gain for live-helper.)

Cody Somerville says that some degree of parallelisation speeds up live filesystem builds for OEM Services. We will investigate whether this is also true for livecd-rootfs, and if so what maximum degree of parallelisation we can allow.

We will add a facility to remove the /usr/share/doc report and the fdupes run from release builds.

We will add a dpkg option to turn off I/O safety (fsync et al) and use that during live filesystem builds.

Secondary work

We will evaluate live-helper as a replacement for livecd-rootfs. In order to commit to this, we would at minimum need to do a file-by-file comparison of the output and sign off on any differences; a moderate amount of customisation work will almost certainly be required.

Test/Demo Plan

It's important that we are able to test new features, and demonstrate them to users. Use this section to describe a short plan that anybody can follow that demonstrates the feature is working. This can then be used during testing, and to show off after release. Please add an entry to http://testcases.qa.ubuntu.com/Coverage/NewFeatures for tracking test coverage.

This need not be added or completed until the specification is nearing beta.


CategorySpec

FoundationsTeam/Specs/MaverickCDBuildSpeed (last edited 2010-09-29 18:28:59 by 99-156-85-10)