[AURORA-1617] Install instructions should point out the critical step of matching mesos slave --work_dir to the observer --mesos-root

• Type: Task
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: 0.13.0
• Component/s: Documentation, Observer
• Assignee: John Sirois
• Reporter: John Sirois
• Created: 12/Feb/16 17:22
• Updated: 12/Apr/16 13:58
• Resolved: 12/Feb/16 19:43

Description

As reported by Thorhs here: http://wilderness.apache.org/channels/?f=aurora/2016-02-12#1455288155

Fri Feb 12 14:44:09 2016  	Thorhs:	Hi All, I'm trying out the nightly build of aurora, and things are going fine until I try to view the task state in the observer. Clicking in the Host link in the Active Task page redirects me to port 1338/task/taskid, but it gives me a 404 error.
Fri Feb 12 14:44:30 2016  	Thorhs:	I think it may be mismatch between the checkpoint path between the executor and the observer.
Fri Feb 12 14:44:39 2016  	Thorhs:	Is this something that you have seen before?
Fri Feb 12 14:46:31 2016  	Thorhs:	I'm running on Centos7 with RPM aurora-scheduler-0.13.0snapshot.2016.02.10-1.el7.centos.aurora.x86_64
Fri Feb 12 14:53:22 2016  	Thorhs:	Looking at the thermos, it appears it is looking for a directory with checkpoints, defaulting to /var/run/thermos if I read it correctly. on the ps output for an executor, I see the checkpoint path is set to /tmp/mesos/slaves/886fc9bc-179b-43c4-a7c6-e706ab7ae96b-S0/frameworks/20160210-072614-3231125002-5050-1392-0000/executors/thermos-1455287988358-nobody-devel-hello_world-0-9fe076da-1037-447c-95e8-ff8ca7751834/runs/a3efc6bd-d108-49
Fri Feb 12 14:58:35 2016  	igmor:	Joined the channel
Fri Feb 12 15:08:04 2016  	Thorhs:	Never mind, work_dir was not set in /etc/mesos-slave. Once set to /var/lib/mesos and restarted, everything started working. I must have missed a step in the instructions.