launchpad-buildd

launchpad buildd abort does not work as expected

Bug #54730 reported by Celso Providelo
20
This bug affects 2 people
Affects Status Importance Assigned to Milestone
Launchpad itself
Fix Released
High
Colin Watson
launchpad-buildd
Fix Released
High
Colin Watson

Bug Description

Most of the times "ABORT" xmlrpc command fails to kill the process, we are probably not sending the correct signal. It works if you kill child process by hand (with 15)

Tags: qa-ok

Related branches

lp:~cjwatson/launchpad-buildd/fix-abort
William Grant: Approve (code)
Adam Conrad (community): Approve
Diff: 1160 lines (+562/-180)
14 files modified
Makefile (+1/-0)
debian/changelog (+7/-0)
debian/upgrade-config (+19/-1)
lpbuildd/binarypackage.py (+9/-6)
lpbuildd/debian.py (+34/-20)
lpbuildd/slave.py (+123/-48)
lpbuildd/sourcepackagerecipe.py (+6/-2)
lpbuildd/tests/buildd-slave-test.conf (+1/-2)
lpbuildd/tests/fakeslave.py (+104/-0)
lpbuildd/tests/harness.py (+1/-0)
lpbuildd/tests/test_binarypackage.py (+206/-0)
lpbuildd/tests/test_translationtemplatesbuildmanager.py (+44/-96)
lpbuildd/translationtemplates.py (+6/-4)
template-buildd-slave.conf (+1/-1)
lp:~cjwatson/launchpad/buildstatus-aborted
William Grant: Approve (code)
Diff: 325 lines (+124/-66)
8 files modified
lib/lp/buildmaster/interfaces/builder.py (+15/-6)
lib/lp/buildmaster/manager.py (+5/-20)
lib/lp/buildmaster/model/builder.py (+27/-9)
lib/lp/buildmaster/model/buildfarmjobbehavior.py (+26/-1)
lib/lp/buildmaster/tests/test_builder.py (+8/-2)
lib/lp/buildmaster/tests/test_buildfarmjobbehavior.py (+29/-3)
lib/lp/buildmaster/tests/test_manager.py (+0/-25)
lib/lp/soyuz/tests/test_binarypackagebuildbehavior.py (+14/-0)
lp:~cjwatson/launchpad/buildmaster-cancel-properly
William Grant: Approve (code)
Diff: 424 lines (+144/-98)
6 files modified
lib/lp/buildmaster/interactor.py (+0/-10)
lib/lp/buildmaster/manager.py (+50/-17)
lib/lp/buildmaster/tests/mock_slaves.py (+5/-1)
lib/lp/buildmaster/tests/test_builder.py (+9/-13)
lib/lp/buildmaster/tests/test_manager.py (+79/-54)
lib/lp/soyuz/model/binarypackagebuild.py (+1/-3)
Revision history for this message
Celso Providelo (cprov) wrote :

promoted to high, since it's a lack of functionality supposed to work, however ABORT isn't required very often,

Changed in launchpad-buildd:
importance: Untriaged → High
status: Unconfirmed → Confirmed
Revision history for this message
Adam Conrad (adconrad) wrote :

Fixing this bug depends on fixing #32236 as well, both will be worked on at the same time.

Changed in launchpad-buildd:
assignee: nobody → adconrad
Revision history for this message
Celso Providelo (cprov) wrote :

It doesn't seem that fixing bug #32236 helped with ABORT implementation. So, still pending further investigation.

Changed in launchpad-buildd:
assignee: Adam Conrad (adconrad) → nobody
Revision history for this message
William Grant (wgrant) wrote :

abort() works sometimes, but hangs other times. It caused big problems this morning when lots of builders were automatically rescued, and attempts to abort builds on them hung.

Revision history for this message
William Grant (wgrant) wrote :

Actually, it seems that abort() never works properly. The cases in which I thought it worked were actually just caused by the build finishing.

It seems to me that it could never have worked, since there has always been a shell script in between the slave and sbuild (at least as far as I can see back in RF). abort() will attempt to and succeed in killing sbuild-package, but that will leave sbuild running. sbuild inherits sbuild-package's FDs, so processEnded isn't called until sbuild dies too.

If I alter sbuild-package to trap SIGTERM and forward it on to sbuild, abort() works fine!

Revision history for this message
Julian Edwards (julian-edwards) wrote :

\o/

Revision history for this message
Robert Collins (lifeless) wrote :

So, this is an sbuild-package bug.

Revision history for this message
Robert Collins (lifeless) wrote :

I wish there were a way to mark the launchpad build system task 'blocked' or something.

Revision history for this message
Julian Edwards (julian-edwards) wrote :

William, is it fixed in the newer sbuild that you were working on upgrading us to?

Revision history for this message
William Grant (wgrant) wrote :

sbuild-package calls sbuild; it's part of launchpad-buildd.

Changed in sbuild (Ubuntu):
status: New → Invalid
Revision history for this message
Robert Collins (lifeless) wrote : Re: [Bug 54730] Re: launchpad buildd abort does not work as expected

Ok, so you've got a patch then?

Revision history for this message
William Grant (wgrant) wrote :

I did, but it's lost in ancient history somewhere. I'll recreate it next week.

Revision history for this message
Robert Collins (lifeless) wrote :

@wgrant ping

tags: added: easy
no longer affects: sbuild (Ubuntu)
Colin Watson (cjwatson)
Changed in launchpad-buildd:
assignee: nobody → Colin Watson (cjwatson)
status: Triaged → In Progress
Colin Watson (cjwatson)
tags: removed: easy
Revision history for this message
Launchpad QA Bot (lpqabot) wrote :

Fixed in stable r16727 <http://bazaar.launchpad.net/~launchpad-pqm/launchpad/stable/revision/16727>.

tags: added: qa-needstesting
Changed in launchpad-buildd:
status: In Progress → Fix Committed
William Grant (wgrant)
tags: added: qa-ok
removed: qa-needstesting
Revision history for this message
Launchpad QA Bot (lpqabot) wrote :

Fixed in stable r16748 <http://bazaar.launchpad.net/~launchpad-pqm/launchpad/stable/revision/16748>.

tags: added: qa-needstesting
removed: qa-ok
Colin Watson (cjwatson)
tags: added: qa-ok
removed: qa-needstesting
Colin Watson (cjwatson)
Changed in launchpad:
status: New → Fix Committed
importance: Undecided → High
assignee: nobody → Colin Watson (cjwatson)
Changed in launchpad-buildd:
status: Fix Committed → Fix Released
William Grant (wgrant)
Changed in launchpad:
status: Fix Committed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Subscribing...

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.