SSD Space Limits about compute-image-tools

[zmarano]

Yes we use PD-SSD's for performance reasons (it is significantly faster for this particular workflow). We recommend you up your PD-SSD quota in the region you are in. Quota increases are free and since these are short lived resources the cost is minimal. https://cloud.google.com/compute/quotas

We will also take a look at allowing non SSD disks for this purpose with the disclaimer that the time to export can be dramatically increased.

[CartWill]

Thank you. We’re using the free trial right now to evaluate if we’ll move to GCP. The quota is stuck at 100GB until we’d upgrade to a regular account so that’s good to know we can request an increase.

[PeterHindes]

The problem is that the program allocates an insanely large ssd without regard to the image size. This degrades the experience of free-trial users.

[zoran15]

Setting export_instance_disk_size variable when running Daisy workflow overrides SSD size in question. This SSD doesn't store image being exported and can be as low as 10GB if quotas are an issue.

[CartWill]

No the problem was there is a hard 100GB limit on the SSD size quota for free-tier users. As I remember it, you can set the size but you cannot set the type. Even though I wasn't using a SSD in my machine, the workflow uses them behind the scenes to facilitate the export process. The workaround is really to request an increase however free-tier trial users can't request increases until they become paying customers. If we had decided to use GCP, maybe we could have forked the workflow and done some kind of customization to the script to not use SSD, however that's more effort we were willing to put in on this while were were just on the free evaluation tier. Ultimately, we decided on a different platform rather than continue trying to get GCP to work for our needs.