Task took Swarming bot offline |
||||||
Issue descriptionThe following task: https://chromium-swarm.appspot.com/user/task/3120d46edf385910 took build545-m4 offline: https://chromium-swarm.appspot.com/restricted/bot/build545-m4 Strangely, the task ID that the bot reports as its last one is different than this one (the low bit is set; not sure whether that means something in Swarming), though it appears to be the same task: https://chromium-swarm.appspot.com/user/task/3120d46edf385911 This was from this CL: https://codereview.chromium.org/2320023002 I filed Issue 645279 about bringing up this and some other dead bots. I'm not sure why the WindowServer is being killed by these jobs, but we need to make Swarming more resilient to this failure mode.
,
Sep 16 2016
Note: d4d44f9ccbce0cd089a3066c438952863921cd40 is intended to work around a graphics driver bug that was affecting this test. It just landed today, so one of the most recent instances of this problem: https://chromium-swarm.appspot.com/user/task/3146977600a55010 happened before it landed. Let's continue to watch chromium_try_flakes for this problem ( Issue 619264 ) and see if it's resolved: https://chromium-try-flakes.appspot.com/all_flake_occurrences?key=ahVzfmNocm9taXVtLXRyeS1mbGFrZXNyMAsSBUZsYWtlIiV3ZWJnbDJfY29uZm9ybWFuY2VfdGVzdHMgKHdpdGggcGF0Y2gpDA Swarming should be resilient to this failure mode though, and reboot the machine if it happens.
,
Sep 23 2016
-RVG, there's nothing internal. Will take another look at this specifically on the Swarming side.
,
Nov 8 2016
Issue 619264 appears to continue to be flaking, but should this issue be closed at this point?
,
Nov 9 2016
We can downgrade this to P2 (or lower) at this point, but no changes went into the Swarming code in response yet, and I think some should in order to make it more robust.
,
Oct 30 2017
Some general improvement went into the bot; but I'm not familiar enough with OSX to know how to detect this state. As long as we don't know what killed the bot, I can't determine what to do to detect the error state.
,
Nov 21 2017
Is this still a problem worth investigating? |
||||||
►
Sign in to add a comment |
||||||
Comment 1 by kbr@chromium.org
, Sep 14 2016