CentOS 6 (x86_64) - with Updates HVM
CentOS.org | 2002_01Linux/Unix, CentOS 6 - 64-bit Amazon Machine Image (AMI)
ssh using private key still prompts for password
Recently I've been using ami-bc8131d4 for lab purposes.
After I've built a custom image with Packer using above ami as a source and then created an instance using Terraform I couldn't connect as root with associated key pair. I've been asked for password.
I got it working by injecting 'PasswordAuthentication no' to sshd_config and public key to authorized_keys while building my image.
Anybody had an issue like that and knows the reason of this behaviour?
- Leave a Comment |
- Mark review as helpful
Clean and with cloud-init support
I was struggling getting a shell script in EC2 user-data to work, when I realized I was using deprecated official CentOS AMIs that was lacking cloud-init support. I upgraded to this one (thanks to the official documentation: https://wiki.centos.org/Cloud/AWS) and I'm happy to see it is already included.
So, OK, maybe root ssh login is disabled, but nothing prevents you from getting it back using a small user-data script:
#!/bin/bash
sed -i -e 's/^.*\(ssh.*$\)/\1/' /root/.ssh/authorized_keys
login fails for root on the latest version
root login seems to be disbled on 2015-09-28 version.
however root login works on 2014-09-28 version.
Hopefully CentOS team take a note of this
by the moment work as well
Is certain that the max default volume is 8GB, but applying the second short guide of NCBI - SRA was solved.
Thanks
Root disk is always 8GB!
The only thing that stinks big time with this image is that root filesystem is always 8GB even though you allocate more while launching an instance. I don't think it is safe to "resize" a live filesystem so you need to use a dirty workaround to resize the filesystem image. Ubuntu and some other images don't have this problem.
Root volume unexpandable
I was a bad experience using this IAM. I needed a root volume larger than 8 GB. Even though I changed the size at the time of instance creation, it didn't workout. I tried to expand the volume later and that did not work too. later I found in online forums that the root volume is not expandable because of the way IAM was created. It wasted lot of my time and it was a horrible experience. If the root volume doesn't expand it must made clear instead of wasting customers valuable time and money.
In frustration, I wanted to take my business elsewhere but things worked out using another IAM.
Please make sure next time for the IAMs the behavious and specifications are mentoned.
Regards,
Mk
No surprises
I wanted CentOS 6, and that is exactly what I got. No tricks, nothing special, just CentOS 6. It comes up as CentOS 6.5, which you can update to 6.6 in a couple of minutes (and then snapshot). I just used it as a way of testing Drupal 7 before running it locally. Easy.
Not "Cloud Ready"
This is specifically about the AMI, not the CentOS 6.6 product itself. It unfortunately suffers from the following faults:
1) 8GiB fixed root (/) partition
2) No cloud-init
3) Clumsy root ssh key handling - instead of using cloud-init, it uses /etc/rc.local in a while loop to check for the SSH key. You can leave an old key in there very easily.
4) As of this writing (late June 2015) it has not been made compatible with the t2.large & c4.* type instances.
The first three points could actually be fixed by installing cloud-init and configuring it. For my organization I've rebuilt the AMI with:
# yum install -y cloud-init
# yum install -y dracut-modules-growpart cloud-utils-growpart
# to /etc/cloud/cloud.cfg, add:
growpart:
mode: auto
devices: ['/']
ignore_growroot_disabled: false
resize_rootfs: True
# dracut -f
Then create a new AMI. On boot, the cloud-init "growpart" module will grow the root partition to the total EBS size. Clearly, if that's not what you what, change or ignore that advice.
You can also fix this manually with a repartition and a reboot, but this will not automatically work as a new image.
# fdisk /dev/xvda <<END
u
d
n
p
1
2048
w
END
# reboot
# (after reboot)
# resize2fs /dev/xvda1
# df -h / # should now be entire EBS volume
Disk setup really sucks
I don't understand all those who say that this image comes with LVM.
I instantiated an new VM this moment just to be sure - US-EAST-1, ami-c2a818aa - no LVM tools, and when installing LVM tools manually (`yum install lvm2`), there are no volume groups detected.
Resizing the disk using fdisk on the live root works, but requires a reboot and feels wrong.
I would recreate the image myself, but I can't attach the drive as a slave to another vm for cloning, and can't recreate the AMI in my Xen setup because the build script for this AMI is not publicly available (or can found, anyway).
Otherwise it's no better or worse than the PV image.
CentOS 6 (x86_64) - with Updates HVM
Hi,
We are really happy to say this image, which is more convenient for us. Then we are expecting lvm image from the Centos end.