[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] getting an email when a job goes on hold?



Todd Tannenbaum <tannenba@xxxxxxxxxxx> writes:
> Daniel Pittman wrote:
>> G'day.
>>
>> One of my users, today, asked if it was possible to get an email when a job is
>> placed on hold due to a failure of some sort?  They are using the job options
>> to cause it to remain in the queue at that point.
>>
>> As far as I can see this isn't presently supported; is that correct?
>
> The good news:
> I believe you will get email if you put in your submit description file
>   notification = always
>   notify_user = email_address@xxxxxxxxxxxx

I will annoy the user in question by telling them this, since the other email
is undesirable. ;)

> The bad news:
> With notification=always, however, you'll also get email lots of other times
> (when job run, completes, etc).  It'd be nice if you got email when/if the job
> is placed on hold if
>   notification = error
> but I am not sure that will always happen because I think Condor's
> interpretation of what is an "error" right now is very strict (aka just if the
> job dies abnormally w/ a signal and/or a handful of unrecoverable errors
> attempting to launch the job like bad permissions).
>
> What do others think? Should we make a change so that the user always gets
> email if the job goes in hold when notification=error?

I am not sure that "error" includes on hold in my head, since that can happen
for reasons other than errors (periodic hold, presumably, isn't an error).

I can see a notification state that includes error and also hold, for a
"notify when not running", being useful?

        Daniel
-- 
✣ Daniel Pittman            ✉ daniel@xxxxxxxxxxxx            ☎ +61 401 155 707
               ♽ made with 100 percent post-consumer electrons