Details
Description
Hi,
My driver application is running in a container (it is a metronome job). It invokes a spark SQL request. Sometimes (not systematic), the execution fails. In mesos GUI (Graphical User Interface), framework, completed tasks, all tasks (7 tasks) are shown as KILLED.
In master server, using "journalctl -u dcos-mesos-master -b | less", I can see:
Jan 04 10:48:56 versailles-bcmt-master-2.ca4mn.com mesos-master[11092]: I0104 10:48:56.281100 11107 master.cpp:1284] Framework fc494a1f-479d-42f2-a2b8-350d383f86bd-1119 (Summarization) at scheduler-08001282-f999-4af3-a7ad-7de2f7da222c@10.75.219.13:45282 disconnected
In attachment, traces.log.gz is output for:
journalctl -u dcos-mesos-master -b | grep "Jan 04 10" > traces.log