partition_count [SQL0094]
Edit this pageAs it appears on __system.alerts:
partition_count
A stream’s message log holds enough partitions that dropping the stream approaches the lock-table ceiling.
The hourly check counts every stream’s message-log partitions and publishes
an active alert for each stream at or above the job’s threshold. Dropping or
destroying the stream locks about 5 relations per partition in one
transaction, and past max_locks_per_transaction Postgres refuses it with
out of shared memory. The alert’s data carries partition_count,
lock_ceiling, and threshold.
Lower the stream’s retention so the janitor drops old partitions, or raise
max_locks_per_transaction. The next run that finds the count under the
threshold publishes the resolution under the same key.
sqlstreams alert listLooking for a different code? Search every thread.