Amazon EC2 micro instance not responding

Question

I have a couple of micro instances that have been working fine for weeks. Both are running WordPress blogs. In the last 24 hours one of them has stopped. I can't ssh in even after a reboot. The other instance is working fine.

ssh: connect to host ec2-xxx-xxx-xxx-xxx.ap-southeast-1.compute.amazonaws.com port 22: Operation timed out

There in nothing obvious in the log that looks like a problem. The last few lines are:

cloud-init:  runcmd[  OK  ]
Mounting other filesystems:  [  OK  ]
Retrigger failed udev events[  OK  ]
Generating SSH1 RSA host key: [  OK  ]
Starting sshd: [  OK  ]
Starting ntpd: [  OK  ]
Starting sendmail: [  OK  ]
Starting sm-client: [  OK  ]
Starting crond: [  OK  ]
[  OK  ]
Starting atd: [  OK  ]
Starting yum-updatesd: [  OK  ]
Running cloud-init user-scripts (none found)[  OK  ]
Amazon Linux AMI release 2011.02.1.1 (beta)
Kernel 2.6.35.11-83.9.amzn1.i686 on an i686
ip-xx-xxx-xx-xx login:

The management console states that everything is running and normal.
I use the same security group and .pem file for both instances.

I suspect that this instance has been getting more traffic than the other one. Is there anyway that the micro instance could run out of memory and just stop responding? What could be going wrong?

Here is a screen shot of the Monitoring panel

Thanks

Answer 1

I've seen micro instances lock up for several minutes due to the CPU "stealing" that occurs when you use too much CPU. This is unique to the micro instance. I blogged an example of this (including video) at http://gregsramblings.com/2011/02/07/amazon-ec2-micro-instance-cpu-steal/.

You can move your instance to new resources simply by doing a full STOP and then a START. This will assign it to new hardware and will assign a new IP address (don't forget to re-associate your elastic IP!). A host reboot will not accomplish this. It needs to be stopped via the EC2 console. Terminating it is not necessary.

Answer 2

There are several possibilities, but the two most likely are:

1. High load on the host that your Micro instance is running on - Micro instances get a small slice of resources anyway, and get scaled back quite harshly when the host is under load.

2. A fault has occurred on the host which is impacting VM responsiveness - this is actually relatively common, and can exhibit the type of behaviour you're seeing.

In either case, the quickest solution is to nuke the instance and restart it - you'll likely get a new instance on a different host, which may be less stressed or less broken. ;)