Closed Bug 535920 Opened 15 years ago Closed 15 years ago

Create updates for 3.0.16 build2 to 3.0.17

Categories

(Release Engineering :: General, defect, P3)

defect

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: coop, Assigned: nthomas)

References

Details

(Whiteboard: updates done, still want to move firefox/releases/3.0.16-real to 3.0.16)

There was a snafu late during the release process for 3.0.16 where build2 got pushed to mirrors instead of build3. We should offer an update for any users who may have managed to grab the build2 installer before we noticed the error. Perhaps the first question is: do we know how many users are affected? Is it enough for us to bother? Second: do we move the files from 3.0.16-real to 3.0.16? What's the procedure here? Copy files, change bouncer entries, remove old files? Third, never having had to do this myself before, what's the procedure for generating this update? Here's my general overview based on the MU I had to create for 3.0.16, but this could be way off: * update patcher config (do we make a new one? modify existing moz19-branch-patcher2.cfg?) * update release/updates/moz19-firefox-* * build patcher tools (via patcher) * download mars (via patcher) * figure out what directory structure needs to be for patcher and create symlinks * create partial patches (via patcher) * quick verify * push to AUS * run backupsnip + pushnip (are we creating test+beta+release snippets)?
Blocks: 531813
(In reply to comment #0) > Perhaps the first question is: do we know how many users are affected? Is it > enough for us to bother? The metrics website says there were 300 users with win32 3.0.16 build2 hitting AUS yesterday, and 1 linux or mac. Surprisingly few. > Second: do we move the files from 3.0.16-real to 3.0.16? What's the procedure > here? Copy files, change bouncer entries, remove old files? Ideally we would copy from 3.0.16-real/ to 3.0.16/, set up some temporary bouncer entries to track uptake, update the main bouncer entries once we had sufficent coverage, remove 3.0.16-real/ and update the mozilla-current rsync module. Unfortunately there is a small subset of mirrors that rsync without --delete, so they still have build2 in 3.0.16/. Bouncer won't exclude those mirrors automatically so we need to reach out to them to catch up. Dave, do you have contacts for these ? http://hyperion.zih.tu-dresden.de/moz http://mozilla.mirror.pacific.net.au/firefox/releases/ http://mozilla.cdn.cacheboy.net/ (tried pinging adrian on irc) That's much improved over a few days ago when there more than 10. These are enabled mirrors but currently non-responsive, so we need to check if/when they come back (or disable them): http://mirror.uoregon.edu/mozilla/releases http://mozilla.initrd.net http://mozilla.prunk.si http://less.cogeco.net/ftp/mozilla-releases http://mozilla.isohunt.com (only very recently down, usually a stalwart) http://ftp.cgu.edu.tw/Mirror/Mozilla/ http://ups.df.lth.se/mozilla > Third, never having had to do this myself before, what's the procedure for > generating this update? We just want to give the build2 users a complete to 3.0.17 I think. Some snippet copying should get us to glory quite quickly (eg from 3.0.15).
(In reply to comment #1) > > Third, never having had to do this myself before, what's the procedure for > > generating this update? > > We just want to give the build2 users a complete to 3.0.17 I think. Some > snippet copying should get us to glory quite quickly (eg from 3.0.15). Yes, especially given the quick turnaround on 3.0.17, I guess it makes more sense to just update from 3.0.16 build2 -> 3.0.17 rather than to 3.0.16 build3. Changing dependency and summary.
Blocks: 535790
No longer blocks: 531813
Summary: Create updates for 3.0.16 build2 to build3 → Create updates for 3.0.16 build2 to 3.0.17
All these users will be on the release channel so we can deal with this in January.
Assignee: nobody → nrthomas
Priority: -- → P2
# aus2-staging cd /opt/aus2/snippets/staging mkdir 20100110-Firefox-3.0.16build1-build2-fillin cd 20100110-Firefox-3.0.16build1-build2-fillin mkdir -p Firefox/3.0.16/{Darwin_Universal-gcc3,Linux_x86-gcc3,WINNT_x86-msvc} cd Firefox/ # build1, build2 pairs ln -s ../../3.0.15/Darwin_Universal-gcc3/2009101600 \ 3.0.16/Darwin_Universal-gcc3/2009113010 ln -s ../../3.0.15/Darwin_Universal-gcc3/2009101600 \ 3.0.16/Darwin_Universal-gcc3/2009113019 ln -s ../../3.0.15/Linux_x86-gcc3/2009101600 3.0.16/Linux_x86-gcc3/2009113010 ln -s ../../3.0.15/Linux_x86-gcc3/2009101600 3.0.16/Linux_x86-gcc3/2009113019 ln -s ../../3.0.15/WINNT_x86-msvc/2009101601 3.0.16/WINNT_x86-msvc/2009113011 ln -s ../../3.0.15/WINNT_x86-msvc/2009101601 3.0.16/WINNT_x86-msvc/2009113020 ~/bin/pushsnip 20100110-Firefox-3.0.16build1-build2-fillin Verified working using Linux builds and some url querying.
These active mirrors still have a 3.0.16 directory: http://hyperion.zih.tu-dresden.de/moz http://mozilla.mirror.pacific.net.au/ These ones are now fine: http://mozilla.cdn.cacheboy.net/ http://mirror.uoregon.edu/mozilla/releases These ones are still down: http://mozilla.initrd.net http://mozilla.prunk.si http://less.cogeco.net/ftp/mozilla-releases http://mozilla.isohunt.com http://ftp.cgu.edu.tw/Mirror/Mozilla/ http://ups.df.lth.se/mozilla I've added a Firefox-3.0.16-Test product to Bouncer, pointing at the incorrect installers in firefox/releases/3.0.16/, to make it easier to track which mirrors are stale.
isohunt.com is down because of router issues, they're doing a manual sync now to make sure they're up to date when returning.
Status: * 3.0.16 build2 to 3.0.17 updates available since 10 Jan. * Trying to get hold of administrators at hyperion.zih.tu-dresden.du and mozilla.mirror.pacific.net.au to get them to remove firefox/releases/3.0.16 and use rsync --delete.
Priority: P2 → P3
Whiteboard: updates done, still want to move firefox/releases/3.0.16-real to 3.0.16
No response. 3.0.16-real it will have to stay.
Status: NEW → RESOLVED
Closed: 15 years ago
Resolution: --- → FIXED
Product: mozilla.org → Release Engineering
You need to log in before you can comment on or make changes to this bug.