Releases: HumanCellAtlas/lira
Releases · HumanCellAtlas/lira
v0.11.0
- Add "v6" subscription queries for SmartSeq2 and 10x data
- Include the fields in notification metadata attachments as workflow labels
- Set
max_cromwell_retries
in workflow inputs (defaults to 0) - Support HMAC auth for notifications
- Fix url for registering service account in FireCloud.
- Move dss_url, ingest_url and Cromwell-as-a-Service params into Lira config for greater flexibility.
- Hard code kube cluster name to "listener" in lira-deploy Jenkins job to prevent error when multiple clusters are present.
- Require both user and password when not using Cromwell-as-a-Service
- Improve instructions for updating a certificate
v0.10.0
- Improves the scalability and stability of Lira by:
- Increasing the number of Gunicorn workers.
- Increasing the timeout and graceful timeout of Gunicorn workers.
- Replacing the sync workers with
gevent
workers.
- Adds the option to include additional metadata in notifications to the utility script when it is making subscriptions.
- Enables Lira to on hold workflows when it is submitting to Cromwell.
v0.9.1
Update kube-deploy script to pass in USE_CAAS
parameter when creating the listener-deployment.yaml file.
v0.9.0
Add version endpoint to Lira that includes information about its configuration (e.g. what pipeline versions it is running)
v0.8.0
- Add cromwell-as-a-service support to lira
- Freeze libraries in requirements.txt
v0.7.0
v0.6.2
- Dynamically add environment-specific urls for the HCA data storage service and integration service into the lira config
- Make it possible to use a stub submission wdl
v0.6.1
v0.6.0
Changes:
-
Cromwell-as-a-Service support
There are new scripts in the repo which help set up the required user accounts for lira to talk to Cromwell-as-a-Service: -
Logging
Lira now can send logs to the HCA DCP's centralized logging system. There are also some bug fixes to Lira's logging mechanism.
v0.5.0
Changes:
- Python 3 support
Lira now works in Python 3 as well as Python 2. Previously it worked in Python 2 only. - WDL caching
Lira now caches WDLs instead of downloading them anew for every workflow launch. Caching only works in Python 3. In Python 2 it still downloads the WDLs prior to each workflow launch. Caching is enabled by default but can be turned off by adding"cache_wdls": false
to Lira's config. - Dry run mode
For testing purposes, Lira can now run in "dry run" mode, by addingdry_run: true
to Lira's config. In this mode, Lira prepares submissions to Cromwell upon receiving notifications but does not actually send the workflows to Cromwell. - Config field renamed
wdl_default_inputs_link has been renamed wdl_static_inputs_link. Existing configurations must be modified to use the new field name or will cause an error on startup.