Releases: iExecBlockchainComputing/iexec-core
Releases · iExecBlockchainComputing/iexec-core
v7.3.0
v7.2.2
v7.2.1
- #533 Replace
sessionIdimplementation with a hash of the public configuration. From a consumer point of view, a constant hash received from thePOST /pingresponse indicates that the scheduler configuration has not changed. With such constant hash, either the scheduler has restarted or not, the consumer does not need to reboot.
v7.2.0
v7.1.3
v7.1.2
v7.1.1
v7.1.0
v7.0.1
What's changed?
- #483 Reduce probability of giving tasks whose consensus is already reached to additional workers.
- #481 Remove useless logs.
- #480 Trigger contribution deadline detector safely.
- #479 #468 Handle task supply and task update management based on their level of priority.
- #473 Upgrade automated build system.
v7.0.0 - The 7th Gear
Highly improved throughput of the iExec protocol.
What has changed since v6.0.0?
- #475 Fix task status deadlock. Chances to reach RUNNING task status for given states of replicates are now increased.
- #465 Fix race condition on replicate attribution.
- #469 Upgrade Jacoco/Sonarqube reporting and plugins.
- #457 #461 Consume blockchain configuration on iexec-blockchain-adapter-api & expose its URL for iexec-worker
- #464 Upgrade artifacts publishing
- #463 Enable local import of iexec-common
- #459 Upgrade to Gradle 6.8.3
- #462 Upgrade to JUnit5
- #454 Fix concurrent upload requests.
- #455 Merge abort notifications into a single notification with custom abort cause.
- #418 Send transactions over a dedicated blockchain adapter micro-service.
- #420 Reply gracefully to the worker when a status is already reported.
- #427 Abort a TEE task when all alive workers did fail to run it.
- #436 Move task state machine to a dedicated service.
- #440 Remove useless internal task statuses belonging to the result upload request stage.
- #446 Fix
Recoveringfor theRetryableupdateReplicateStatus(..)method. - #444 Add checks before locally upgrading to the
INITIALIZEDstatus - #444 Remove 2-blocks waiting time before supplying a new replicate
- #442 Fix
OptimisticLockingFailureExceptionhappening when 2 detectors detect the same change at the same time, leading to race updates on a same task - #437 Reuse socket when sending multiple requests to a blockchain node.
- #425 Replay fromBlockNumber now lives in a dedicated configuration:
- A configuration document did store two different states. Updates on different states at the same time might lead to race conditions when saving to database. Now each state has its own document to avoid race conditions when saving.
- #423 Fix
TaskRepositoy.findByChainDealIdAndTaskIndex()to return unique result