Job Manager release notesRSS

Improvements

• Adds upgraded logging functionality to include exceptions when they are thrown.

Fixes

• Skipped querying for the instanceId if the Metadata field is disabled to avoid Server Side Request Forgery attacks that can leak information

Improvements

• Largely improved the reduction in compute resources due to more efficient usage of each job manager
• Added more detailed logging for when a job not found
• Provided better logging around jobs coming off the parking log

Fixes

• Increased container memory to 3.256 Gib and shm size to 2.256 Gib to allow Chrome to have enough resources
• Updated dropwizard to version 2.0.34 to remediate Text4Shell Vulnerability (CVE-2022-42889)

New Features

• Changed the MAX_JOB_CAPACITY environment variable to HEAVYWEIGHT_WORKERS environment variable, which is the maximum number of heavyweight jobs that the job manager will run in parallel

Improvements

• Updated the synthetics-utilities version we are using to 1.0.24

Fixes

• Added better logging around the error message for not finding the runtimes
• Fixed an NPE error about the ping-runtime could not be found after all the runtimes have been downloaded successfully
• Replaced docker cpu quotas with docker cpu shares to ensure that unused capacity is not wasted

New Features

• Added the ability to configure a CPU core limit for each runtime type

Improvements

• Reduced CPU usage while the job manager is waiting for results from ephemeral runtimes by implementing a more efficient waiting algorithm

Improvements

• Improved AWS error messages

Fixes

• Fixed a bug where the PrivateRuntimeDownloadUtility wasn't evaluating the default registry correctly

New Features

• Adding DESIRED_RUNTIMES as an environment variable, so users can supply their particularly desired runtime(s) and tags

Improvements

• Making DOCKER_REGISTRY and DOCKER_REPOSITORY configurable via environment variables to replace the default docker.io and newrelic values, respectively

Fixes

• Sending all events data via the environment variable rather than just through US_EAST_1

New Features

• Adding MAX_JOB_CAPACITY, which is the maximum number of heavyweight jobs that the job manager will run in parallel, as an environment variable

Improvements

• Added a pending queue for private job managers, before getting jobs from Horde