zhongxingzhi created MESOS-784:
----------------------------------
Summary: slave should allocate all the rest resources even given a
resources flag
Key: MESOS-784
URL: https://issues.apache.org/jira/browse/MESOS-784
Project: Mesos
Issue Type: Bug
Components: slave
Affects Versions: 0.16.0
Environment: rh 6.3
Reporter: zhongxingzhi
Fix For: 0.16.0
I use two computers for testing.
pc1: cpu:6, mem:8G, disk:20G.
pc2: cpu:8, mem:10G, disk:30G.
first, start the master,
mesos-master --roles="role1"
and then start each slave with --resources flag like that:
mesos-slave --resources = "mem(role1):1024";
after slave connect to master, you would see each slave has only 1024M mem, the
reset is invisible. Maybe add the rest mem to each --resources flag Manually
will be a good idea (--resouces = "mem(role1):1024;mem(*):7000").
A shell script will be used when you has many slaves with different mem,
another problem appear:
mesos-slave --resources = "mem(role1):10000000000000", it start with no error,
but when the slave launch tasks, the slave crash.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
