Closed Bug 648207 Opened 13 years ago Closed 13 years ago

do staging runs of 5.0b1 and 5.0b2

Categories

(Release Engineering :: General, defect)

x86_64
Linux
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: bhearsum, Assigned: bhearsum)

Details

In a meeting about the new release process today it was brought up that drivers would like to do releases from mozilla-beta with the in-app version set to a final version, eg. "5.0". However, we'll be doing many releases with the same version, all of which need to get updates to future releases, and be pushed to mirrors in a unique directory.

I think we can use the existing version/appVersion functionality built into the release automation for this, but we need to do a couple of staging runs to verify. Specifically, we should do one with version=5.0b1, and one with 5.0b2.
I just finished doing these staging runs, and I believe we're in good shape to run releases out of mozilla-beta that are internally versioned as "5.0", and still enable us to properly serve updates to them, and push them to mirrors.

To be clear, all of the following apply to these builds:
- The internal version used for blocklist, AUS, and AMO pings is "5.0"
- The title bar, about window, installed icons, install directory, and all other branding says "Firefox" or "Mozilla Firefox" -- there is no "Beta N" appended.
- When pushed to mirrors, these go to a "5.0bN" directory.
- The filenames used also contain "5.0bN", eg:
http://releases.mozilla.org/pub/mozilla.org/firefox/releases/5.0b1/win32/en-US/Firefox Setup 5.0 Beta 1.exe
- Bouncer links will use "5.0bN", too. Eg, http://download.mozilla.org/?product=firefox-5.0b1&os=win&lang=en-US

Christian and others who care -- are all of the above okay?
(In reply to comment #1)
> - The internal version used for blocklist, AUS, and AMO pings is "5.0"

Related question - do we have any bugs on improving metrics and crash-stats to be able to tell the difference between these N.0 builds ? I think we're going to want to be able to identify them and those groups currently rely on version to do that. eg How crashy is the latest build ? How many of our users are on the latest build ? 

I know that metrics can query down to the buildid if you ask them, and advanced search on crash-stats can do buildID, but I think we might want to go further than that.
Metrics and the Socorro team are aware of the change, not sure if any work is scoped or bugs are written.
(In reply to comment #3)
> Metrics and the Socorro team are aware of the change, not sure if any work is
> scoped or bugs are written.

Any response to the bullet points in comment #1?
(In reply to comment #1)
> I just finished doing these staging runs, and I believe we're in good shape to
> run releases out of mozilla-beta that are internally versioned as "5.0", and
> still enable us to properly serve updates to them, and push them to mirrors.
> 
> To be clear, all of the following apply to these builds:
> - The internal version used for blocklist, AUS, and AMO pings is "5.0"

Cool.

> - The title bar, about window, installed icons, install directory, and all
> other branding says "Firefox" or "Mozilla Firefox" -- there is no "Beta N"
> appended.

Cool.

> - When pushed to mirrors, these go to a "5.0bN" directory.

Cool.

> - The filenames used also contain "5.0bN", eg:
> http://releases.mozilla.org/pub/mozilla.org/firefox/releases/5.0b1/win32/en-US/Firefox
> Setup 5.0 Beta 1.exe

That's fine and it matches the current method, correct? Ideally they don't have the beta bits in the filename (as it is already in the dir) but it isn't a big deal either way.

> - Bouncer links will use "5.0bN", too. Eg,
> http://download.mozilla.org/?product=firefox-5.0b1&os=win&lang=en-US
> 

That's fine for now.
AFAIK, we have build IDs in metrics and crash-stats, I have no idea about channels on the metrics side, but I hear that crash-stats doesn't have them as breakpad doesn't deliver them.

If we want channels in addition to version and build ID to be in crashes, we need to file bugs on both Breakpad Integration and Socorro and get implementations fast, I guess. Do we need that?
(In reply to comment #5)
> (In reply to comment #1)
> > - The filenames used also contain "5.0bN", eg:
> > http://releases.mozilla.org/pub/mozilla.org/firefox/releases/5.0b1/win32/en-US/Firefox
> > Setup 5.0 Beta 1.exe
> 
> That's fine and it matches the current method, correct? Ideally they don't have
> the beta bits in the filename (as it is already in the dir) but it isn't a big
> deal either way.

I'm not sure how much work it'll be to change that, so let's stick with what we have for now. We can revisit later.


Since everything seems OK, I'm marking this as FIXED and declaring RelEng ready to go from an automation standpoint, woohoo!
Status: ASSIGNED → RESOLVED
Closed: 13 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.