TODO

[ ] stream logs
[ ] check benchmark execution mlcube training exp ID (what is this? I forgot)
[ ] double check if we need atomic transactions on the server
[x] ~~We now have demo data url and hash in training exp (dummy) that we don't use.~~
[ ] test a common scenario: model owner and agg owner being the same
[x] ~~collaborators doesn't use tensorboard logs.~~ (related to logging, first point)
[x] ~~finalize endpoints for new entities? (delete?, put?, ...)~~
[ ] have timeouts in integration tests (3mins should be enough)
[x] docker publish port only to a specified network interface
[x] ~~limit mlcube network access? (for now we can rely on the review of the experiment owner)~~
[ ] server permission: they are broken!!
[x] plan being confidential (related to permissions
[x] running datasets before aggs??
[x] mounting files: be careful!!! check parent folders (issue will be created)
[ ] 404 for exp.aggregator,ca,event: not informative (check common practice for such a thing)
[ ] writing unit tests
[ ] finalize CA issues/deployment

Apr 30 '24 13:04 hasan7n

FL support