Date Jul 03, 2014
Attendees
Discussion Items
Steve
- Not stuck; VMs are good
- Working on having worker contact Archive DMCS
- Passing hostname and port at job start
- Could also query DNS or have other mechanisms (even configuration)
- Using socket message; could use DB or HTTP/REST
Greg
- Scenario with app-level failed worker
- DAG runs, exits, leaves rescue DAG
- Doesn't rerun because it thinks it won't get better
- But there is support for automatic retry as option in DAG
- Adding that now; looks like it works
- Will work on summarizing runs into logs/video/diagram