2013-06-19

Spot instances that stop (or reboot) on their own: a new feature in AWS ?

So I have a spot instance for my visualization project. I am using one t1.micro instance in the spot market.

My instance is i-05cabc6a and its spot request is sir-4e01da35. It has a 64 GB EBS volume attached to it (vol-0eba4a7f) and a 8 GB EBS volume for the operating system (vol-b23d42e9). 

According to the EC2 Management Console, its launch time was 2013-06-06 11:00 EDT (314 hours). This is a spot instance, it can not be stopped. It can only be terminated, or obviously it can continue to live too.


So today I was showing people my visualization project, but it did not work.

I use http://browser.cloud.raytrek.com/client for the permanent address.

In the DNS:

;; ANSWER SECTION:
browser.cloud.raytrek.com. 14113 IN     CNAME   browser-10.raytrek.com.
browser-10.raytrek.com. 14113   IN      CNAME   ec2-54-226-13-244.compute-1.amazonaws.com.
ec2-54-226-13-244.compute-1.amazonaws.com. 155620 IN A 54.226.13.244

So there was something wrong. I decided from this point to connect to my instance.

I was surprised to see that the uptime of the operating system on my instance was only 1 day and 20 hours. Today is 2013-06-19, and my spot instance started on 2013-06-06.

[web@ip-10-125-9-5 ~]$ w
 17:52:05 up 1 day, 20:08,  1 user,  load average: 0.00, 0.01, 0.05
USER     TTY      FROM              LOGIN@   IDLE   JCPU   PCPU WHAT
ec2-user pts/0    132.203.106.196  17:45    5.00s  0.03s  0.01s sshd: ec2-user [priv]
 

My web server was not running neither.

[web@ip-10-125-9-5 ~]$ ps aux|grep lighttpd|grep -v grep

This the second time that this is happening to me, see my Twitter. A Senior Product Manager from Amazon Web Services even contacted me about this in the past. See the message below (I redacted names and addresses):


Hi Sébastien,

I’m the Sr. Product Manager focusing on Amazon EC2 Spot Instances and I noticed your tweet that your Spot instance was stopped instead of terminated. We definitely didn’t introduce a new feature—I’d love to understand what happened. Would you be willing to share some details to help us figure it out (e.g., Spot instance request id)?

Thanks!
############

########### | Amazon Web Services—EC2 | Sr. Product Manager | #############| ###########

In my case, starting my stuff takes less than 5 seconds since all my data and software are in Amazon EBS.

More commands to start Ray Cloud Browser (on TCP port 80):

[ec2-user@ip-10-125-9-5 ~]$ sudo su
[root@ip-10-125-9-5 ec2-user]# cd /mnt/
[root@ip-10-125-9-5 mnt]# ls
vol-0a34c87b  x
[root@ip-10-125-9-5 mnt]# mkdir vol-0eba4a7f
[root@ip-10-125-9-5 mnt]# mount /dev/sdf vol-0eba4a7f/
[root@ip-10-125-9-5 mnt]# exit
[ec2-user@ip-10-125-9-5 ~]$ sudo su web
find: failed to restore initial working directory: Permission denied
[web@ip-10-125-9-5 ec2-user]$ cd
[web@ip-10-125-9-5 ~]$ ls
Ray-Cloud-Browser-instance
[web@ip-10-125-9-5 ~]$ rm -rf Ray-Cloud-Browser-instance
[web@ip-10-125-9-5 ~]$ ln -s /mnt/vol-0eba4a7f/
lost+found/       vol-0a34c87b.tar  web/             
[web@ip-10-125-9-5 ~]$ ln -s /mnt/vol-0eba4a7f/web/Ray-Cloud-Browser-instance
[web@ip-10-125-9-5 ~]$ ls
Ray-Cloud-Browser-instance
[web@ip-10-125-9-5 ~]$ cd Ray-Cloud-Browser-instance/
[web@ip-10-125-9-5 Ray-Cloud-Browser-instance]$ ls
lighttpd.conf  logs  pid.txt  Ray-Cloud-Browser  Ray-Cloud-Browser-data
[web@ip-10-125-9-5 Ray-Cloud-Browser-instance]$ exit
[ec2-user@ip-10-125-9-5 ~]$ sudo su
[root@ip-10-125-9-5 ec2-user]# cd /home/web/Ray-Cloud-Browser-instance/
[root@ip-10-125-9-5 Ray-Cloud-Browser-instance]# lighttpd -f lighttpd.conf

1 comment:

Blogger said...

Bluehost is the best web-hosting provider with plans for all of your hosting needs.

There was an error in this gadget