61Every write to the cluster fails and etcd reports a NOSPACE alarm. The disk has plenty of room. Get us writing again.▼hardNewGoogleRed HatDatabricks◆ premiumThe disk is not full, the keyspace quota is. Defragmenting without disarming the alarm leaves the cluster read-only, and disarming without compacting puts it back within minutes.Open full answer →
09Six teams share a GPU cluster. Two of them are idle most nights. Design the quota model.▼hardNewDatabricksUberMicrosoft2 repliesunlockedHard quotas strand idle GPUs and soft quotas mean the loudest team owns the cluster. Borrowing with reclaim is the shape that works, and the interesting decisions are all about what happens when the lender comes back.Open full answer →
