sandbox launch fails with "access denied" |
||||
Issue descriptionOS: Windows BrokerServices::SpawnTarget occasionally fails with last_error set to 5 (access denied). In the field this happens more often on Windows 7. An internal test can reproduce this on Windows 7 by spawning about 40 sandbox subprocesses that do minimal work in MULTIPROCESS_TEST_MAIN. The failures don't happen with --single-process-tests but it does happen with LaunchUnitTestsSerially (which I undertand splits into multiple shards but runs the tests in serial within each shard). Is it possible that with sharding starting so many short-lived sandboxes is running out of system resources?
,
Aug 3
I have a unit test that's failing repeatedly due to this error on win7 now. Sample failure: https://ci.chromium.org/p/chromium/builders/luci.chromium.try/win7_chromium_rel_ng/53472 This is the first time a sandbox has been started in the test run, so it doesn't seem to be due to load. Bumping to P2 since this is blocking open-sourcing the Chrome Cleanup tool - this unit test is pretty basic that I don't want to continue with the open sourcing until it's working.
,
Nov 21
,
Jan 11
This issue has an owner, a component and a priority, but is still listed as untriaged or unconfirmed. By definition, this bug is triaged. Changing status to "assigned". Please reach out to me if you disagree with how I've done this. |
||||
►
Sign in to add a comment |
||||
Comment 1 by joenotcharles@chromium.org
, Jun 26 2018