Skip to content

0.7.3: fix(cluster): prevent premature closure when a new model is deployed

Choose a tag to compare

@abde-crunchdao abde-crunchdao released this 15 Oct 10:57
· 21 commits to master since this release
When a new model was deployed replacing the previous one, if the WS "RUNNING" message from the new model arrived before the old model was replaced, the cluster incorrectly closed the new instance.