If the worker is alive it will handle the failure (and send the email). If the worker goes offline and stops emitting heartbeats, the schedule will eventually handle the failure and send the email.
On Wed, Jan 10, 2018 at 9:48 AM, Sakshi Barnwal <[email protected]> wrote: > Hi Max, > > Yeah, so, we have certain port issues on the worker node, hence, I was > doubtful that in a generic scenario, since the scheduler passes the task to > the worker node, what happens if the task fails, the mail config (smtp > sender, port) will be picked from worker or master. > > Thanks, > Sakshi > > On Wed, Jan 10, 2018 at 11:15 PM, Maxime Beauchemin < > [email protected]> wrote: > > > Most people use an identical airflow.cfg across all machines on the > Airflow > > cluster. > > > > Max > > > > On Wed, Jan 10, 2018 at 9:31 AM, Sakshi Barnwal < > [email protected] > > > > > wrote: > > > > > My doubt was airflow config of which machine will be picked up, the > > master > > > (scheduler ) or worker. > > > > > > > > > On Wed, Jan 10, 2018 at 8:52 PM, manish ranjan <[email protected]> > > > wrote: > > > > > > > SMTP configurations go to airflow.config. > > > > the email list is part of every dag, in default_args { 'email': > > [email protected], > > > > [email protected]} > > > > > > > > ~Manish > > > > > > > > > > > > > > > > On Wed, Jan 10, 2018 at 5:13 AM, Sakshi Barnwal < > > > [email protected] > > > > > > > > > wrote: > > > > > > > > > Hi folks, > > > > > > > > > > I have a doubt with respect to working of Airflow. > > > > > > > > > > We have a two node cluster (scheduler, webserver and another for > > > worker). > > > > > So, when a task or dag has failed or is retried, and a mail is sent > > to > > > > the > > > > > concerned list, which configuration file will be taken into > account , > > > for > > > > > sending the retry/failure mail. > > > > > > > > > > > > > > > Thanks, > > > > > Sakshi > > > > > > > > > > > > > > >
