FreeBSD on Amazon EC2

jdereus said:
Here's a new Dutch provider of FreeBSD 8.1 systems: http://www.tilaa.nl

I'm a happy customer :)

FreeBSD doesn't have paravirtualized network and block drivers for KVM, which means FreeBSD has less than optimal performance. In practise it works quite well, but if high performance is a requirement for your application we currently recommend a Linux based OS.
There appears to be a regression in KVM with regard to ACPI events. For some reason the ACPI power button event is not received by the FreeBSD guest. This effectively means you can't gracefully shutdown your FreeBSD VPS from our webapp. We expect this issue to be fixed soon-ish.
We can't resize the FreeBSD UFS filesystem from the Linux host, nor does the FreeBSD growfs command support online resizing. This makes it difficult if not impossible to properly automate resizing of FreeBSD guests. If you want to upgrade your FreeBSD VPS later on, be aware that the filesystems will not be resized. (Our webapp will give a warning about this.) If you want to start using the extra assigned disk space, contact us and we'll grow your filesystem manually.

Too bad they don't sell jails. Instead it's FreeBSD on KVM :\
 
You can ask me ;) for a Jail. I hope that I can start up with my Startup on the first January. Currently I have to decide me between NetBSD with XEN, Scientific Linux with XEN or a pure FreeBSD with VIMAGE. They will be boxes with two Opterons, up to 64 GB Memory and fast SAS drives.
 
Running FreeBSD on Amazon EC2

Information:
http://www.daemonology.net/blog/2010-12-13-FreeBSD-on-EC2.html
http://www.daemonology.net/freebsd-on-ec2/

AMI: ami-c01aeca9

To Connect:
1. Once you have started an instance, click on your running instance in the console and copy the Public DNS to your clipboard.
2. Execute the following command, making substitutions for items in bold:

Code:
ssh -i [i][b]path to your .pem file[/i][/b] root@[i][b]Your Public DNS Name[/b][/i]

3. Remember: this is only for 9.0 CURRENT and still in the experimental stages, so deployment of production systems is not recommended.
 
  • Thanks
Reactions: pez
Anyone Using The EC2?

Is anyone else using the EC2 micro instances? I have been, with much success, and am happy to see that there is now an 8.2-RELEASE AMI. Thanks, Colin!

I was trying to add some EBS storage today and ran into some trouble. My instance crashed and dumped core. I'm wondering if anyone else has been able to add extra storage, beyond the 10G that are available in the AMIs?

Thanks.
 
EC2: Adding EBS Disks To A Running Instance

Going to answer my own question here.

Colin emailed me with the following info:

This is a bug in fdisk. There is a simple workaround: Don't use fdisk --
there's really no point given that you can create as many virtual disks
as you want using EBS. Just newfs and mount the raw EBS disks.

So, once you create an EBS Volume and attach it to your running instance, tail your /var/log/messages and you will see the device name assigned by the kernel:

Code:
Jun 10 15:55:37 yourhost kernel: xbd2: 10240MB <Virtual Block Device> at device/vbd/2080 on xenbusb_front0
Jun 10 15:55:37 yourhost kernel: xbd2: attaching as da2
Jun 10 15:55:37 yourhost kernel: GEOM: new disk da2

..in this case, it's /dev/da2, So...

Code:
# newfs /dev/da2
# mkdir /ebs
# mount /dev/da2 /ebs

...now you have new storage you can access at /ebs.

Don't forget to add an entry in /etc/fstab if you want to mount it at boot:

Code:
/dev/da2 /ebs ufs rw 0 2
 
I am using Micro Instances a lot for all the small tasks you don't need or don't want to have on your productive machines. Micro Instances - at least in my case - also do a great job as loadbalancing reverse proxies.

An attached EBS harddisk cannot be seen like a physical HDD. You can use it just like any RAW block device.
RAID works great actually - tried RAID0 and got a real performance boost surprisingly. Should use RAID5 or 10 / 01 for mission critical data of course.

Tip: EBS has some sort of lazy physical space allocation mechanism which (extremely) slows down the first write on each virtual block.

dd if=/dev/zero the complete disk before you are using it for the first time takes ages but helps big time.
 
Amazon AWS EC2 FreeBSD Micro images (micro only) have a bug related to the fstab file. Seems that when you edit fstab and reboot the virtual server then it does not complete reboot. It hangs due to panic over the edited entry in the fstab that related to newly attached volumes. Since reboot stops when it panics the IP does not come up. Users need to launce a new instance as a work around... which is a large pain in the rear.
 
Is there an ETA for AMIs in the new Oregon (us-west-2) region? Is there a way I can move one from us-west-1? Any info on this would be appreciated.
 
for the record, I was able to migrate a machine to a different region using dump and restore, using these steps:

1. Attach and mount a separate disk from your root disk
2. Stop services, and do a dump backup of your root dir to a file on the other disk
3. Start an instance of the same type in the remote region (New)
4. Start another insance of the same type in the remote region (Temp)
5. Copy the dump file from the original machine to the Temp instance
6. Shutdown (Stop) the New instance.
7. Detach the main drive from the new instance and attach it to the Temp instance.
8. On the Temp instance, mount the boot drive from the New instance.
9. On the Temp instance, cd into the mount point for the New instance's boot drive
10. Restore the dump file into the root of the New instance's drive
11. Unmount and and detach the New instance's boot drive from the Temp instance.
12. Re-attach the New instance's boot drive to the New instance. Type: /dev/sda1 as the device name.
13. Boot the new instance.
 
Back
Top