Closed
Bug 1012518
Opened 11 years ago
Closed 10 years ago
Figure out retention of flame tinderbox-builds
Categories
(Release Engineering :: General, defect)
Tracking
(Not tracked)
RESOLVED
FIXED
People
(Reporter: nthomas, Assigned: nthomas)
References
Details
Attachments
(2 files)
2.32 KB,
patch
|
Callek
:
review+
nthomas
:
checked-in+
|
Details | Diff | Splinter Review |
868 bytes,
patch
|
nthomas
:
checked-in-
|
Details | Diff | Splinter Review |
Like bug 969767, but for flame. Do we know what we want to do yet Clint ?
Flags: needinfo?(ctalbert)
Ideally, I want to keep them all. :-)
The flame is going to be our reference device for the next year. At the very least we should keep the 12-18 weeks as this will give us the ability to bisect across the lifetime of the 2.0 release using these builds.
As we saw with 1.3, 1.3t, and 1.4 the releases live long after the 12 week "release cycle". So, we need to keep at least 12 weeks worth of builds. That's our minimum.
Also, now that we have a phone with a sane flashing story, we might be able to do some build on the fly and flash bisection tools so that we don't necessarily have to keep the tinderbox builds around. But it will take a few weeks to get that effort together.
So right now, we want at least the 12 weeks of flame device builds kept. More if you can get it. :-)
Thanks
Flags: needinfo?(ctalbert)
Assignee | ||
Comment 3•10 years ago
|
||
In order of importance:
* add 12 week retention for all flame builds. This is much simpler than white/backlisting branches, and the contribution from older gaia and project branches is tiny compared to b2g-i and m-i anyway
* limit 12 week retention for hamachi to mozilla-b2g30_v1_4 branch
* stop deleting crashsymbols & logs, they are a small contribution
(< 5%) compared to device images and gecko/gaia tarballs
* while we're here, removes cron on empty mnt/pvt_builds/pvt/mozilla.org/b2g/tinderbox-builds
Projecting from current disk usage we will use an additional 2.5T of space, which is within the current allocation.
Attachment #8465782 -
Flags: review?(bugspam.Callek)
Comment 4•10 years ago
|
||
Comment on attachment 8465782 [details] [diff] [review]
[svn] cron changes
Review of attachment 8465782 [details] [diff] [review]:
-----------------------------------------------------------------
pretty much just stamping, but based on your explanation this all sounds great
Attachment #8465782 -
Flags: review?(bugspam.Callek) → review+
Assignee | ||
Comment 5•10 years ago
|
||
Comment on attachment 8465782 [details] [diff] [review]
[svn] cron changes
Committed revision 91279.
Attachment #8465782 -
Flags: checked-in+
Assignee | ||
Comment 6•10 years ago
|
||
Leaving open for verification early next week.
Assignee | ||
Comment 7•10 years ago
|
||
The change from 21 to 7 days for emulator builds was not intended, reverted:
Committed revision 91348.
Attachment #8466839 -
Flags: checked-in+
Assignee | ||
Comment 8•10 years ago
|
||
Verified working. I expedited the cleanup of dirs like b2g-inbound-hamachi*, the previous style of removing crash symbols and logs meant they had mtimes quite a bit later than when the builds were produced.
Status: ASSIGNED → RESOLVED
Closed: 10 years ago
Resolution: --- → FIXED
Assignee | ||
Comment 9•10 years ago
|
||
Comment on attachment 8466839 [details] [diff] [review]
[svn] Partial revert
Turns out I misread my earlier patch and had never changed it in the first place (pub vs pvt), and we're running filling up the disk now. I've backed it out.
Committed revision 91741.
Attachment #8466839 -
Flags: checked-in+ → checked-in-
You need to log in
before you can comment on or make changes to this bug.
Description
•