Bug #17511

website cache should be keyed on debian/changelog

Added by intrigeri 2020-03-09 07:34:27 . Updated 2020-03-10 17:36:13 .

Status:
Resolved
Priority:
Elevated
Assignee:
CyrilBrulebois
Category:
Build system
Target version:
Start date:
Due date:
% Done:

0%

Feature Branch:
bugfix/17511-website-cache-vs-differing-changelog-entries
Type of work:
Code
Blueprint:

Starter:
Affected tool:
Deliverable for:

Description

Otherwise, for example we can hit this scenario:

  1. during a built from the stable branch, the website is built with SOURCE_DATE_EPOCH derived from the last changelog entry on that branch (date1); the result of that build is cached
  2. during another build from the devel branch, on a different isobuilder, the website is built with SOURCE_DATE_EPOCH derived from the last changelog entry on that branch (date2); the result of that build is cached
  3. when Jenkins tries to reproduce build 2, on the same isobuilder as build 1, our build system reuses the website built with date1, which does not match the website built with date2 during build 2 ⇒ different images, reproducibility check fails

Example: https://jenkins.tails.boum.org/view/RM/job/reproducibly_build_Tails_ISO_devel/933/artifact/build-artifacts/diffoscope.iso.html


Subtasks


History

#1 Updated by intrigeri 2020-03-09 07:42:49

  • Status changed from Confirmed to In Progress

Applied in changeset commit:tails|792cb70ac395944372f2328e183a4a402c8dc5df.

#2 Updated by intrigeri 2020-03-09 07:43:26

  • Feature Branch set to bugfix/17511-website-cache-vs-differing-changelog-entries

#3 Updated by intrigeri 2020-03-09 09:43:48

  • Status changed from In Progress to Needs Validation
  • Assignee changed from intrigeri to CyrilBrulebois

Hello @CyrilBrulebois, this should fix the reproducibility problem we’ve been discussing on the RM mailing list today! I thought you might want to take this in 4.4.

#4 Updated by CyrilBrulebois 2020-03-10 04:17:42

Oh wow, I’m glad I asked the question…

Merged, thanks!

#5 Updated by CyrilBrulebois 2020-03-10 04:18:40

  • Status changed from Needs Validation to Resolved

#6 Updated by intrigeri 2020-03-10 08:41:40

> Oh wow, I’m glad I asked the question…

Same here :)

#7 Updated by CyrilBrulebois 2020-03-10 17:36:13

And in case anyone wonders: having just updated debian/changelog by expanding on the latest changes (without changing the date), I can confirm the website cache was invalidated as expected… :)