New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 711185 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Closed: Apr 2017
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug



Sign in to add a comment

No space left on vm1074-m4

Project Member Reported by tikuta@chromium.org, Apr 13 2017

Issue description

https://luci-milo.appspot.com/buildbot/tryserver.chromium.mac/ios-simulator/193391

Compile step failed.

[20326/20485] ACTION //ios/chrome/test/earl_grey:ios_chrome_settings_egtests_executable(//build/toolchain/mac:ios_clang_x64)
FAILED: obj/ios/chrome/test/earl_grey/ios_chrome_settings_egtests 
python ../../build/toolchain/mac/linker_driver.py --developer_dir /b/build/slave/ios-simulator/build/src/build/ios_files/Xcode.app xcrun lipo -create -output obj/ios/chrome/test/earl_grey/ios_chrome_settings_egtests obj/ios/chrome/test/earl_grey/x64/ios_chrome_settings_egtests ios_clang_x86/obj/ios/chrome/test/earl_grey/x86/ios_chrome_settings_egtests
fatal error: /b/build/slave/ios-simulator/build/src/build/ios_files/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/lipo: can't write to output file: obj/ios/chrome/test/earl_grey/ios_chrome_settings_egtests.lipo (No space left on device)
Traceback (most recent call last):
  File "../../build/toolchain/mac/linker_driver.py", line 229, in <module>
    Main(sys.argv)
  File "../../build/toolchain/mac/linker_driver.py", line 79, in Main
    subprocess.check_call(compiler_driver_args)
  File "/System/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/subprocess.py", line 540, in check_call
    raise CalledProcessError(retcode, cmd)
subprocess.CalledProcessError: Command '['xcrun', 'lipo', '-create', '-output', 'obj/ios/chrome/test/earl_grey/ios_chrome_settings_egtests', 'obj/ios/chrome/test/earl_grey/x64/ios_chrome_settings_egtests', 'ios_clang_x86/obj/ios/chrome/test/earl_grey/x86/ios_chrome_settings_egtests']' returned non-zero exit status 1

 

Comment 1 by tikuta@chromium.org, Apr 13 2017

Labels: Infra-Troopers

Comment 2 by tikuta@chromium.org, Apr 13 2017

Labels: -Hotlist-Infra-Troopers

Comment 3 by d...@chromium.org, Apr 13 2017

Cc: sergeybe...@chromium.org smut@chromium.org
Owner: d...@chromium.org
Status: Fixed (was: Untriaged)
This builder had "build/build.dead" and "build/ios_simulator" directories, both of which had old timestamps (March 21). I'm deleting both of them.

+smut@, +sergeyberezin@, this is the second time I hit this. Is this representative if a pandemic? Do we need to add these directories to a cleanup script?

Comment 4 by s...@google.com, Apr 13 2017

I thought build.dead was supposed to be removed by something. Can we add it to cleanup_disk if it's not there already?

Comment 5 by d...@chromium.org, Apr 13 2017

/b/build/slave/build.dead is removed. This one was /b/build/slave/build/build.dead. I think someone manually created that one and put in the wrong place :P
According to https://chrome-internal.googlesource.com/infra/infra_internal/+/master/go/src/infra_internal/tools/cleanup_disk/cmd/cleanup_disk/main.go#103

we remove /b/build/slave/*/build.dead, so the file you deleted should've been deleted automatically. However, cleanup_disk runs only once a day around midnight PST, so if build.dead appeared recently and caused the disk to fill up, there isn't much we can do automatically at the moment.

But in general, if you see problematic files appear at scale, consider adding a glob to the above config and redeploy cleanup_disk (through puppet).

Comment 7 by d...@chromium.org, Apr 13 2017

/b/build/slave/*/build.dead
               ^---- one star = one directory substitution

This isn't the one I'm worried about. I'm fairly certain this was a human putting a thing in the wrong place.

I'm more concerned about build/ios_simulator. Maybe it gets installed once and never touched again? But since I've deleted it, several builds have run and it hasn't been recreated.

Comment 9 by d...@chromium.org, Apr 13 2017

So you think this is residual from that? OK so I wasn't too far off in proposing a pandemic situation :P

Comment 10 by s...@google.com, Apr 13 2017

I think so. Dates coincide, and duplicated checkouts in <buildername> directories seems like the same problem. I guess nobody went through and deleted all the duped checkouts.

https://chrome-internal.googlesource.com/infra/puppet#Emergency-disk-cleanup-on-many-bots

Sign in to add a comment