For all-pages locking in a nonclustered edition of Adaptive Server, if the server fails to acquire a page lock, it retries the attempt instead of waiting for the lock (which can cause a deadlock). The deadlock retries configuration parameter determines the number of retries Adaptive Server attempts.
The Cluster Edition includes a physical lock, which ensures page consistency across nodes. However, a physical lock is not included in the deadlock detection mechanism, so a server cannot detect a deadlock.
The Cluster Edition retries failed attempts to acquire a lock. However, when the server exceeds the value for deadlock retries, it fails or rolls back the query instead of waiting on the lock and risking an undetected deadlock.
