Issue metadata
Sign in to add a comment
|
Bot with state 'need_reset' still run tasks with 'dut_state:ready'. |
||||||||||||||||||||||||
Issue descriptionFor a given bot: https://chrome-swarming.appspot.com/bot?id=chromeos-skylab-bot-46c6c402-c6cb-4b7c-a96d-40f548f364ba&sort_stats=total%3Adesc Here list some of the example failed tests: https://chrome-swarming.appspot.com/task?id=3e322a3cd05c1a10&refresh=10 https://chrome-swarming.appspot.com/task?id=3e322a7a79630510&refresh=10 https://chrome-swarming.appspot.com/task?id=3e322a7dc2d71310&refresh=10 I see 2 issues and guess they're related: 1) The bot should be state of 'need_reset' after the first failure. However, it's still accepting tasks. This makes me assume that these tasks are pre-allocated to this bot. 2) The bot runs different tasks at the same time. Not sure whether it's the reason of the failure 'Client job got aborted.'. Will let @pprabhu to decide whether this should be fixed from swarming side or lucifer.
,
Jun 19 2018
,
Jun 19 2018
Create Issue 854352 for issue 2 . Make this bug focus on a bot's state need_reset should block itself to accept tasks.
,
Jun 20 2018
Most likely this is also a fallout of multiple tasks running on the same bot. Order of events: - Bot is in state:ready - Bot process A picks up test 1 - Bot process B picks up test 2 - test 1 fails. Bot moves to state needs_reset - test 2 succeeds. Bot moves to state ready. - Next test gets scheduled on the bot. |
|||||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||||
Comment 1 by xixuan@chromium.org
, Jun 19 2018