cloud engineer
How to find a boot disk from an instance.
gcloud compute instances describe servername \ --format='get(disks[0].source)' \ --zone=us-central1-c \ --project project-id |
Result
https://www.googleapis.com/compute/v1/projects/project-id/zones/us-central1-f/disks/servername-boot |
How to add a startup-script to metadata. The script executes every time VM is started.
gcloud compute instances add-metadata instance-name \ --project your-project-id \ --zone us-central1-c \ --metadata=startup-script='#! /bin/bash sudo -i echo "Time: $(date)" >> /tmp/date.txt' |
How to resize a root partition in GCP.
Increase disk size.
gcloud compute disks resize server-boot-disk-name \ --size=50GB \ --zone us-central1-c \ --project your-project-id |
Run growpart.
growpart /dev/sda 1 |
Run xfs_growfs.
xfs_growfs /dev/sda1 |
To verify, run these.
lsblk df -Th |
This are grub fixes to servers that are not booting up due to improper entries in the grub menu.
grub --batch <<END root (hd0,0) setup (hd0,0) quit END # sed -i 's/dev\/sda/dev\/hda/' /boot/grub/device.map sed -i 's/(hd0)/(hd0,0)/' /boot/grub/menu.lst |
The fix includes replacing (hd0) with (hd0,0) and /dev/sda with /dev/hda.
Here’s how to find when a Windows server was last rebooted.
systeminfo | find /i "boot time" |
The /etc/fstab file allow volumes to be automatically mounted on bootup. Add nofail in your automated mounts to prevent instances from hanging on boot. Nofail makes the instance to boot even if there are errors with the mounted volume.
UUID=aebf131c-6957-451e-8d34-ec978d9581ae /data xfs defaults,nofail 0 2 |
The following commands will extend Linux Boot drive without a reboot.
List block devices. Check disk sizes.
lsblk df -h |
Determine file system type: xfs, ext2, ext3 or ext4.
file -s /dev/xvda1 # or df -Th |
Extend volume from the console or command line.
aws ec2 modify-volume --region us-east-1 \ --volume-id vol-xxxxxxxxxxxx --size 10 \ --volume-type gp2 |
Resize Linux partition.
# for /dev/xvda1 growpart /dev/xvda 1 # for /dev/xvda2 growpart /dev/xvda 2 # for ext2, ext3, ext4 resize2fs /dev/xvda1 # for xfs xfs_growfs /dev/mapper/root xfs_growfs /dev/xvda1 |
Check if new file size is being displayed.
lsblk df -Th |
Google Cloud Platform just recently released a beta feature called detaching and attaching boot disks. Previously, boot disks were permanently attached to their VM instances. Now you have the ability to detach boot disk from your instance and attach it to another instance without deleting the original instance. You can also replace the boot disks for an instance rather than recreating the entire instance. I’ve tested it on my test account, and it works quite nicely.
I have 2 servers called blue-server and red-server. I’m detaching the disks on blue-server and attaching it to the red-server.
gcloud compute instances detach-disk blue-server --disk=blue-server-disk --zone us-central1-c gcloud compute instances attach-disk cyan-server --disk=blue-server-disk --zone us-central1-c gcloud compute instances detach-disk cyan-server --disk=blue-server-disk --zone us-central1-c gcloud compute instances attach-disk blue-server --disk=blue-server-disk --zone us-central1-c |
Results:
This new GCP feature is still beta. There might be a few quirks here and there, but overall, it’s a must have feature if you are managing OS.