Migrating workloads from digital ocean to gcp

[rimun]

I am in the process of migrating droplets from digital ocean to gcp. Since digital ocean doesn't allow snapshots to be downloaded, I have used third party tools to create full backup of a virtual machine in tar.gz format and uploaded to google cloud storage. I tried to create an image from this using "gcloud compute images create" but it failed with this error: The file inside the tar archive was named ''./''. It should be named ''disk.raw''

I need somebody with prior knowledge to look into this and suggest a workaround to go ahead with this migration.

[dionv]

Hello @rimun ,

For migration of your downloaded tar.gz vm files to Google Cloud Platform, I suggest these steps as per documentation:

1. Extracting the contents of the file. This will likely result in a directory containing the virtual machine's file.
2. Locate the disk image file. Usually the file is disk.img or image.raw file depending on the backup tool used.
3. If the image file is not a raw image, you may consider converting the file. Tools like qemu-img allow format conversion between different disk image formats.
4. Upload the file to Cloud Bucket
5. Create the instance from the image file.

Attaching here the reference documentation for further details.

[Iva]

Another solution: Use a third-party tool directly on GCP

Some third-party cloud backup services offer functionality to directly migrate virtual machines between cloud providers like Digital Ocean, GCP, SharePoint, AWS, Azure....etc. This might eliminate the need for manual extraction and renaming. 

Check out AvePoint, Gs Richcopy 360, Cloudsfer, and GoodSync !