The following release notes correspond to June 18, 2020 - June 24, 2020. In addition to these changes, this release includes back-end updates to workflows, Google integrations, and notebooks to improve upcoming features.
- Workflow status is no longer prevented from updating when a user attempts to overwrite a “built-in” expression such as this.sample_id or this.participant_id with their own data in workflow outputs.
- Now, when you create a new notebook runtime using the standard VM option, you get exactly the disk size you specify.
Previously, when you specified disk size, a portion of that disk was used to install the infrastructure required to run Jupyter notebooks on the machine, and not all disk space was available to you. Going forward, the notebook infrastructure will be installed on a separate boot disk on standard VMs.
Please note that there will be a minor cost increase associated with having a user-specific disk. This minor cost increase is associated with the following improvements:
- Currently, you get exactly the disk size you specify.
- In the near future (1-2 months), this change will enable us to support a detachable persistent disk feature. This ensures that any input data, installed packages, and output data are saved, even when you need to delete/recreate your notebook runtime.
Disk cost example: For a notebook runtime disk, using the default settings with 50 GB of disk space, the disk cost can be compared as follows:
- Previously, the cost for a single disk model (where a portion of the disk is taken up by the notebook infrastructure) would cost $0.0027 per hour, or $2 per month (if you do not delete your notebook runtime and keep the disk for a month).
- Currently, the cost for a separated boot and user disk model (where you get exactly the disk space you specify) would cost $0.0056 per hour, or $4 per month (if you do not delete your notebook runtime and keep the disk for a month).