aurora-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Yuan <yuans4y...@gmail.com>
Subject Re: time constraint on aurora jobs?
Date Thu, 26 Feb 2015 03:13:45 GMT
Thanks Joe. Having a customized StatusChecked is a good way to implement
this time constraint feature. :)

On Tue, Feb 24, 2015 at 11:45 PM, Joseph Smith <yasumoto7@gmail.com> wrote:

> Very good question.. to my knowledge there is not a ‘time’ constraint.
>
> However, you could implement this in a few ways. One of my first thoughts
> is to setup a custom StatusChecker <
> https://github.com/apache/incubator-aurora/blob/e6e7e53d92b52d78960824022bef8a0546002180/src/main/python/apache/aurora/executor/common/status_checker.py#L68>
> which checks the length of a task's runtime. StatusCheckers can return an
> ExitState <
> https://github.com/apache/incubator-aurora/blob/e6e7e53d92b52d78960824022bef8a0546002180/src/main/python/apache/aurora/executor/common/status_checker.py#L27>
> which can end a task. FAILED will allow a Service() to be restarted, but
> KILLED should (if I’m following right) actually prevent that from being
> rescheduled unless a user manually reschedules it, which may or may not be
> what you’re looking for.
>
> An example of this is the HealthChecker <
> https://github.com/apache/incubator-aurora/blob/467bc56049cc775eaf61520a464b363d44023024/src/main/python/apache/aurora/executor/common/health_checker.py>,
> which causes a task to go into ‘FAILED’ if it does not pass a specified
> health check.
>
> Please let me know if that makes sense!
> Joe
>
> > On Feb 24, 2015, at 19:11, Yuan <yuans4yrnn@gmail.com> wrote:
> >
> > Hello,
> >
> >    In apache aurora, there are resource isolations and sizings on CPU,
> > memory and disk space, which can be specified in the job configuration
> > file. Is there any similar way to put a constraint on job running time,
> > like killing a job if it has been running for more than a certain amount
> of
> > time?
> >
> > Thanks,
> > Yuan
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message