1.2.5

• fixed a newly introduced issue of md5 not getting reported in the output.

1.2.3

The newly introduced "use_benchmark" field in "config" can be specified by user.

1.2.2

• New option ebs_throughput available in config of the input job description json. It is applicable only for the gp3 type EBS which is the default.

1.2.1

• Issue of non-admin users not having access to cost_updater step function / executions is now fixed.
• Redeployment is needed for this change to take effect.

1.2.0

• snakemake --tibanna error about not finding snakemake_main_file is now fixed
• By default, 1GB is added to memory and 5GB is added to ebs_size, unless mem_as_is or ebs_size_as_is is set to True in config.
• redeployment is required for these changes to take effect.

1.1.3

tibanna rerun and rerun_many commands now have option --use-spot and --do-not-use-spot

1.1.2

• tag option in output_targets in the input json allow output S3 object tagging as they're created.
• See https://tibanna.readthedocs.io/en/latest/execution_json.html#output-target-specification for more info

1.1.1

• cost estimate now also returns type of the estimate as one of the three:
  • immediate estimate, actual_cost, retrospective estimate
    • The retrospective estimate is the least accurate especially when spot instance was used since the current spot pricing is used rather than the spot pricing at the time of the run. - it will only be used when immediate estimate wasn't available and actual cost expired already
  • If immediate estimate is available, it will not update the cost with retrospective estimate.
• When cost_estimate returns an actual cost, it also fills in the cost field of the metrics report tsv not just cost_estimate, and consistently in the metrics html as well when it is updated.

1.1.0

• Automatic cost updater functionality

  • use -C option when deploying unicorn (you need permission to billing) : this will deploy an extra step function that automatically adds cost information when it becomes available after the run (typically 1~2 days later).
  • tibanna.core API has a new function check_costupdater_status to check the status of the cost updater.

• Dependency can now be specified through job IDs instead of execution ARNs, which allows you to run the next steps before the first step has been submitted because execution ARNs are determined by AWS and is returned once the job gets submitted successfully, whereas job IDs can be optionally specified by the user as part of the run workflow command (-j option). This helps users to set up an automation linking multiple steps more easily.

• encrypt_s3_upload option in config of the input json : when the value it True, it uses KMS encryption for uploading files to a bucket. It applies to every file and every bucket, including output and logs.

• A newly introduced bug in plot_metrics producing an error for failed jobs is fixed.

To use the full features of 1.1.0, redeploy tibanna unicorn (either cleanup and redeploy or deploy a completely new one).

1.0.6

• New function cost_estimate gives an estimated cost while the actual cost is unavailable.
• The cost estimate is auto-calculated at the end of every run.
• Tibanna unicorn must be completely redeployed for it to take effect (not just lambdas) including the IAM role.