I’ve just tried to use Terraform and Packer to create a Mesos + Ceph cluster in AWS. Yes, I know Mesosphere applications supporting deployment of Mesos cluster in some IaaS (see Getting Started), but I’d like to understand what’s going on there. So, I did it by Terraform and Packer. I’m gonna explain a little bit more about this.
Through this work, I’ve learned a lot of things. I will write something below too.
I just tried to run Fluentd instances on Mesos + Docker + Marathon in Vagrant boxes. Here is the sample repository.
The overview diagram:
You may have used chown command to change the ownership of files or directories like below.
$ ls -l file
-rw-r--r-- 1 root root 675 2013-09-23 23:23 file
$ sudo chown foo file
$ ls -l file
-rw-r--r-- 1 foo root 675 2013-09-23 23:23 file
But this is not good because group of file is still root. Most case, you want to set it the default(login) group of the user.
$ id foo
uid=1010(foo) gid=1009(foo) groups=1009(foo)
In this case, you want to set it foo. For that, I used a command below.
$ chown foo:foo file
$ ls -l file
-rw-r--r-- 1 foo foo 675 2013-09-23 23:23 file
Not bad. But you have to know the default group of the user. It is not always same as the user name.
To understand development of Mesos Framework, I developed a simple useless framework example called msh.
In the pervious post, I tried “cpu = 0.1” for Mesos tasks. But it didn’t work precisely. If you have only one cpu on each slave server, you can run only one task on each slave at the same time even if you specify “cpu = 0.1” for the tasks.
It only happens on the slave which has just one cpu. If a slave has more than one cpus, it can run tasks more than the number of cpus. I use m1.small and it has only one cpu, unfortunately…
So, I changed Mesosphere’s runner script a little to configure --resources flag of mesos-slave.
/usr/bin/mesos-init-wrapper
@@ -22,6 +22,7 @@
[[ ! ${ULIMIT:-} ]] || ulimit $ULIMIT
[[ ! ${MASTER:-} ]] || args+=( --master="$MASTER" )
[[ ! ${LOGS:-} ]] || args+=( --log_dir="$LOGS" )
+ [[ ! ${RESOURCES:-} ]] || args+=( --resources="$RESOURCES" )
logged /usr/local/sbin/mesos-slave "${args[@]}"
}
And set more cpus.
$ cat /etc/default/mesos-slave
MASTER=`cat /etc/mesos/zk`
RESOURCES="cpus:2"
Now you can see 2 cpus in your Mesos cluster, then you can run more than 2 tasks on it at the same time.
Today, Mesosphere released an important milestone. I really desired this integration.
Now, we can manage our “Computers as an OS”.
作りたいなーと漠然と思ってるアイデア。自分用メモなので色々省略してますので、興味ある人がいれば詳しく話します。
Typed variable for ruby!
Local host message queue using POSIX mqueue and delayed job. You don’t have to run any daemon for queue store!
foreman written in Go! ddollar is an original author of foreman, so this is a real descendant of foreman.
We can manage go libraries source and version using a simple config file like Bundler.
We can write an desktop application using HTML5, CSS, JS and WebGL!
MongoDB management GUI application. Beautiful…
Today’s my best: node-webkit
Love github!
I love github. I introduce some github repositories from Today’s trends or my recent starred. I will continuously post Github Trends every day hopefully…
I read only README, so my impression must be useless, but I hope it will be a chance for readers to know awesome github repositories.
I tried to use omniauth-openid with my new rails application. It worked when I developed using HTTP. After I started to develop with HTTPS, it failed. Finally I got the coolest workaround: OmniAuth.config.full_host and X-Forwarded-Host/Port.