iExec V6 Middleware: Upgrades, Improvements, Fixes - LIST

[blairmaclennan]

The iExec Core team identified specific areas for improvement for the iExec middleware:

• Websockets: Workers now do not need to be manually whenever there is a communication error between the worker and the scheduler. (GitHub Issues: [#128]), #198, #332, #327)
• Improving Scheduler’s handling of expiring tasks (GitHub Issues: #341, #384)
• Resolved Scheduler State machine bugs (GitHub Issues: #346, #351, #352)
• Fixed worker contribution process bugs (GitHub Issues: #333, #279, #[320]

(#320).)

• Post compute contribution improvement: When the worker encounters an application download problem or a post-compute problem, it cannot contribute. The contribution process will have to be refactored to allow the worker to contribute in as many cases as possible.
• Workers can no longer receive more tasks than their CPUs can handle. (GitHub Issues: #289, #335)
• The SMS now returns the correct HTTP code, even when a problem occurs during a TEE session (GitHub Issue: #15)
• It is now possible to distinguish between log messages produced by the worker and those produced by the running application thanks to a prefix added to the stdout (GitHub Issue: #240)
• Scheduler and worker logs now display the last block number to detect synchronization problems (GitHub Issue #336)
• In the case of the scheduler restarting, workers now wait to complete tasks before restarting themselves (Github Issues: #342)

The iExec V6 technical upgrade includes many more improvements to the iExec middleware and confidential computing solution:

🔒 Enclaves in Production Mode: Intel® SGX hardware enclaves will now be running in full Production Mode! Among other benefits, this makes Public Worker Pools with TEE possible!

🔑 Standard Encryption: Implementing a new standard encryption framework called ‘AES-256-CBC’, it is now easier than ever to create an encrypted dataset in a secure and standardized manner, from any type of device.

Read more about the upgrades here