Closed Bug 387958 Opened 18 years ago Closed 18 years ago

Deploy buildbot patch uploader

Categories

(Release Engineering :: General, defect, P3)

defect

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: bhearsum, Assigned: bhearsum)

References

Details

(Whiteboard: busy with talos stuff)

Attachments

(2 files, 9 obsolete files)

Tracking bug for the deployment of the Buildbot patch uploader.
Hardware: PC → All
Depends on: 387959
Depends on: 387960
Depends on: 387961
Depends on: 387962
Depends on: 387963
Priority: -- → P2
Depends on: 370241
This is a cleaned up version of the master.cfg from test1-linux-buildbot. It has builders for Linux, Mac, and Windows. It's valid according to checkconfig.py.
Attachment #272238 - Flags: review?(rhelmer)
Attached file mozbuild.py needed by master.cfg (obsolete) —
This is the mozbuild.py for master.cfg. The win32-buildbotref-v4 environment _should_ be compatible with the current Win32 Buildbot ref platform.
Attachment #272239 - Flags: review?(rhelmer)
Attachment #272238 - Flags: review?(rhelmer) → review+
Attachment #272239 - Flags: review?(rhelmer) → review+
Attachment #272239 - Attachment description: mozbuild.py needed my master.cfg → mozbuild.py needed by master.cfg
Looks good, let's get these checked in. I suggest: mozilla/tools/buildbot-configs/tryserver/ Any alternative suggestions?
Whiteboard: waiting for checkin
RCS file: /cvsroot/mozilla/tools/buildbot-configs/tryserver/master.cfg,v done Checking in master.cfg; /cvsroot/mozilla/tools/buildbot-configs/tryserver/master.cfg,v <-- master.cfg initial revision: 1.1 done RCS file: /cvsroot/mozilla/tools/buildbot-configs/tryserver/mozbuild.py,v done Checking in mozbuild.py; /cvsroot/mozilla/tools/buildbot-configs/tryserver/mozbuild.py,v <-- mozbuild.py initial revision: 1.1 done
Whiteboard: waiting for checkin → waiting on hardware
The buildmaster is setup on staging-build-console.
Whiteboard: waiting on hardware → waiting on hardware/VMs
Some points that came up on IRC: * How are people going to get results from their submission? Reporting to a Tinderbox or looking at the Waterfall page directly? * How will access be secured? LDAP or despot?
1) the builds should be given some sort of unique ID and then made available for a 24-hour window on some download site 2) Ask IT (reed and I were talking about it briefly)
(In reply to comment #8) > 1) the builds should be given some sort of unique ID and then made available > for a 24-hour window on some download site > You don't think having a Waterfall or Tinderbox display will be useful? I don't think everyone will be downloading builds all the time -- they'll just want to know "does my code burn the tree" or not.
Oh, I kinda figured people would be able to see the buildbot results somewhere, but maybe I shouldn't have made that assumption. So yes, the results should be made available (both build and test results, when we get around to the testing part).
Whiteboard: waiting on hardware/VMs → waiting on new win32 ref platform
As per discussion with joduinn this will be setup with the existing win32 ref platform for now.
Whiteboard: waiting on new win32 ref platform → eta: july 24
Just to be clear: The win32 buildslave will be re-done once the new windows ref platform is available.
Depends on: 389475
Depends on: 388594
No longer depends on: 389475
I'm pushing the ETA back to Friday. Reed says he'll have LDAP accounts for CVS commiters by then. TODO list: * Have the Buildmaster push completed builds back to dm-wwwbuild01 * Authenticate users via LDAP * A bit more testing
Whiteboard: eta: july 24 → eta: july 27
Depends on: 389721
OK. Here's what I've done the past couple days: * Push completed builds (in packages) back to build.m.o * Have the Buildmaster report to the Tinderbox on MozillaTest * Write a BuildStep that ensures only one change per Build is processed * Setup cronjobs on staging-build-console and build.m.o that cleanup patches/binaries older than a week * Run downloadpatches.sh in a cronjob every minute * Add a short note at the top of Tinderbox log so people can easily identify what builds are theirs. Still to do: * Test with many builds queued. * Attach final copies of master.cfg and mozbuild.py to this bug after testing
Attached patch final version of configs (obsolete) — Splinter Review
This is an update to try servers master.cfg and mozbuild.py. Barring any problems these are the configs it will use.
Attachment #272238 - Attachment is obsolete: true
Attachment #272239 - Attachment is obsolete: true
Attachment #274250 - Flags: review?(rhelmer)
Whiteboard: eta: july 27 → in testing
Depends on: 390179
Attachment #274250 - Flags: review?(rhelmer) → review+
Depends on: 390235
No longer depends on: 390235
Attached patch minor tweaks to the last patch (obsolete) — Splinter Review
These are the files that are in use on the tryserver. Can you land these after review?
Attachment #274250 - Attachment is obsolete: true
Attachment #274771 - Flags: review?(rhelmer)
For everyone's information: I moved the master to build-console.b.m.o. I'll be using staging-build-console.b.m.o as a staging area for it.
Noticed this while I was fixing an issue on dm-wwwbuild01: [error] uploadpatch.cgi: Bareword found in conditional at /var/www/html/build/uploadpatch.cgi line 181., referer: https://build.mozilla.org/uploadpatch.cgi
Found the culprit: print PATCH || print("could not upload patch: $!"); The $! is causing the trouble. I'm not sure who put the '|| print ...' in, it's not part of my code.
Depends on: 391368
This version of the configs has support for Mac universal binaries. It should have all the tweaks from your patch too, rhelmer. I'm going to attach the mozconfig files in a minute.
Attachment #274771 - Attachment is obsolete: true
Attachment #275983 - Flags: review?(rhelmer)
Attachment #274771 - Flags: review?(rhelmer)
Attached file tryserver linux mozconfig (obsolete) —
Attachment #275984 - Flags: review?(rhelmer)
Attached file tryserver mac mozconfig (obsolete) —
Attachment #275985 - Flags: review?(rhelmer)
Attached file tryserver win32 mozconfig (obsolete) —
Attachment #275986 - Flags: review?(rhelmer)
Attachment #275983 - Flags: review?(rhelmer) → review+
Attachment #275984 - Flags: review?(rhelmer) → review+
Comment on attachment 275985 [details] tryserver mac mozconfig why does mac have: mk_add_options MOZ_PACKAGE_NSIS=1 ?
Attachment #275986 - Flags: review?(rhelmer) → review+
Attached file tryserver mac mozconfig, fixed (obsolete) —
I must've got mixed up when I was making these.
Attachment #275985 - Attachment is obsolete: true
Attachment #275990 - Flags: review?(rhelmer)
Attachment #275985 - Flags: review?(rhelmer)
Attachment #275990 - Flags: review?(rhelmer) → review+
Depends on: 391354, 392598
Whiteboard: in testing
Depends on: 389422
Priority: P2 → P3
Whiteboard: busy with talos stuff
These are the configs that have been running on the try server for about 2 weeks. I'm confident that they are OK. Sorry to bog you down with another review on this, Rob. Once this is checked in I think we can consider this FIXED.
Attachment #275983 - Attachment is obsolete: true
Attachment #275984 - Attachment is obsolete: true
Attachment #275986 - Attachment is obsolete: true
Attachment #275990 - Attachment is obsolete: true
Attachment #278961 - Flags: review?(rhelmer)
Attachment #278961 - Flags: review?(rhelmer) → review+
Checking in master.cfg; /cvsroot/mozilla/tools/buildbot-configs/tryserver/master.cfg,v <-- master.cfg new revision: 1.2; previous revision: 1.1 done Checking in mozbuild.py; /cvsroot/mozilla/tools/buildbot-configs/tryserver/mozbuild.py,v <-- mozbuild.py new revision: 1.2; previous revision: 1.1 done
Status: ASSIGNED → RESOLVED
Closed: 18 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.

Attachment

General

Created:
Updated:
Size: