You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Now that #310 has been addressed via the pod webhook approach, we need to handle certain race conditions which can occur in this strategy.
Consider the following race condition:
Leader of Job 1 getting killed right after we set the nodeSelector on its follower pods
At the same time, leader of Job 2 lands on the same slice before any of Job 1's followers get to schedule on it. In this case Job 1's followers will have a nodeSelector on the node pool, but will never schedule because of the anti-affinity of Job 2's leader.
The text was updated successfully, but these errors were encountered:
Now that #310 has been addressed via the pod webhook approach, we need to handle certain race conditions which can occur in this strategy.
Consider the following race condition:
The text was updated successfully, but these errors were encountered: