New issue
Advanced search Search tips

Issue 821727 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Closed: Apr 2018
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Feature



Sign in to add a comment

Swarming task page does not provide duration when the bot dies.

Project Member Reported by serg...@chromium.org, Mar 14 2018

Issue description

E.g. see https://chromium-swarm.appspot.com/task?id=3bfff5a075b3ec10. one has to manually calculate the time between 'Started' and 'Abandoned'. Perhaps it can be called differently than 'Duration', but it would be convenient to have that number to see whether bot has come close to some timeout before dying.
 
This affected issue 819237.

Comment 2 by mar...@chromium.org, Mar 14 2018

Labels: Type-Feature
Status: Available (was: Untriaged)
The reason there's no duration is that we may not know the duration in some cases, as when the bot just stopped responding.

But for this case specifically I'll fix the named cache handling.

Comment 3 by mar...@chromium.org, Mar 14 2018

BTW the named cache removal fails because a process outlives the recipe....
Can we have 'Known duration' computed as time between time task started and time bot stopped responding? It's just annoying to have to compute it manually. This is obviously a relatively low-priority feature request.

Comment 5 by maruel@google.com, Mar 16 2018

Yes I agree it should be changed. When a task did run and it's the not that decides to set the internal_failure bit, completed_ts should be used. abandoned_ts should be used when it's the server that kills the build, e.g. if the bot disappeared, like spontaneous reboot.

It's an historical error on my part.
Owner: kjlubick@chromium.org
Status: Fixed (was: Available)
This was fixed in https://chromium-review.googlesource.com/c/infra/luci/luci-py/+/964619

Sign in to add a comment