slave_build path, defined by infra_paths module, is unnecessarily tied
to our buildbot setup which we are trying to deprecate in favor of LUCI.
In practice it is always == start_dir and that's what we want in LUCI too.
Update all code that uses slave_build path to use start_dir instead.
R=iannucci@chromium.org, martiniss@chromium.org
BUG=662586
Change-Id: Id52aa535d6a4f52670d581bf56aab65f670b1861
Reviewed-on: https://chromium-review.googlesource.com/412075
Commit-Queue: Nodir Turakulov <nodir@chromium.org>
Reviewed-by: Stephen Martinis <martiniss@chromium.org>
Remove nested retries on checkout. Since the outer loop handles retries
specially, we want to let it exclusively handle retries. This will
prevent the inner retry loop triggering a timeout condition (rather than
a failure condition) with the outer loop.
BUG=chromium:665116
TEST=None
Change-Id: I4bf0c08b75861bf1c95060452118c314a121f9f3
Reviewed-on: https://chromium-review.googlesource.com/411397
Reviewed-by: Ryan Tseng <hinoka@chromium.org>
Commit-Queue: Ryan Tseng <hinoka@chromium.org>
Currently, 'bot_update' uses the 'gclient' that is on the system path.
Now, it will use the 'gclient.py' that is in the same 'depot_tools'
checkout as the 'bot_update' recipe module.
Also don't ignore "git_cache" move errors.
BUG=664254,663990,663440
TEST=None
Review-Url: https://codereview.chromium.org/2492963002
Hypothesis: Sometimes bot update fails because windows fails to delete
a lockfile associated with a git process.
Test: If this happens, let's delete that lockfile and try again.
BUG=651602
Review-Url: https://codereview.chromium.org/2382653005
Reason for revert:
Pointing to the wrong file :(
Original issue's description:
> bot_update: add --auth-refresh-token-json passthrough for apply_issue
>
> BUG=642150
>
> Committed: e465667e78
TBR=vadimsh@chromium.org,martiniss@chromium.org,tandrii@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=642150
Review-Url: https://codereview.chromium.org/2350363003
Reason for revert:
broke win_chromium_x64_rel_ng/builds/281022
because of apply_issue.bat calls local python.
Original issue's description:
> bot_update: ensure correct depot_tools checkout is used.
>
> R=vadimsh@chromium.org,martiniss@chromium.org
> BUG=635641,647446
>
> Committed: 972ac50401
TBR=martiniss@chromium.org,vadimsh@chromium.org,iannucci@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=635641,647446
Review-Url: https://codereview.chromium.org/2349673002
Reason for revert:
The "git rebase --abort" should only happen when there is a failure during rebase not when it succeeds.
Original issue's description:
> Skip apply_gerrit rebase when there is a rebase failure.
>
> Not doing so causes a failure in the 'git checkout $base_rev' step as visible here: https://build.chromium.org/p/client.skia.fyi/builders/Infra-PerCommit-Trybot/builds/4995
> This causes the checkout to remain in the "rebase in progress" state.
>
>
> BUG=chromium:645955
> BUG=skia:5749
>
> Committed: 9736fe8938
TBR=agable@chromium.org,andybons@chromium.org,tandrii@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:645955
Review-Url: https://codereview.chromium.org/2331103002
This is the simplest change I can make here. The bot_update.py script already
ignores the value of force (and sets it to True in all cases). This is an
expectations-only change.
I'll refactor the content of the bot_update.py script to account for this
removed option in a followup CL as well.
R=agable@chromium.org, estaab@chromium.org, hinoka@chromium.org, martiniss@chromium.org
BUG=643885
Review-Url: https://codereview.chromium.org/2316323002
This makes it so that regardless of what force is set to in the recipe, if you
invoke bot_update.ensure_checkout, it will actually run.
Removal of the force option (and related options) to come shortly.
R=estaab@chromium.org, hinoka@chromium.org, martiniss@chromium.org, tandrii@chromium.org
BUG=643885
Review-Url: https://codereview.chromium.org/2311673002
If an exception occurs during a try block which prevents a step from actually executing, then the finally block's attempt to access step.active_result will fail, and the original exception will be hidden behind a new exception from the recipe engine. This CL changes this behavior so that the original exception will always fire.
Review-Url: https://codereview.chromium.org/2175103002
Reason for revert:
this is wrong. if directory is not empty, git clone fails
Original issue's description:
> bot_update: fix git dir check
>
> existance of a dir does not mean it is a git repo. Check .git dir
> presence instead.
>
> This is the reason why swarmbucket build.git builds fail.
>
> TBR=hinoka@chromium.org
> BUG=
>
> Committed: cdac80c9fbTBR=hinoka@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=
Review-Url: https://codereview.chromium.org/2173683003
site config is not present on builders which use kitchen, so we need to use credentials provided by puppet in the /creds directory.
BUG=624212
TBR=hinoka
Review-Url: https://codereview.chromium.org/2108073002
The script already contains references to chrome-internal, and when running
under remote_run we have no access to hacky build_internal checkout.
BUG=593999, 584079
Review-Url: https://codereview.chromium.org/2101573002
Nodir Turakulov
Dan Jacques
recipe-roller
tandrii
katthomas
borenet
hinoka
vadimsh
rmistry
iannucci
agable
bpastene
andybons
martiniss
smut
phajdan.jr