Large Partitions Hunting¶
This document describes how to catch a huge partition.
Any of the following:
- Latencies on a single shard become very long (look at the “Scylla Overview Metrics” dashboard)
- Huge allocations warning messages in the log:
seastar_memory - oversized allocation: 2842624 bytes, please report
- SSTable are bigger than expected (look at the data directory of the corresponding column family)
- For each table you suspect run:
nodetool flush <keyspace name> <table name> nodetool cfstats <keyspace name>.<table name> | grep "Compacted partition maximum bytes"
nodetool cfstats demodb.tmcr | grep "Compacted partition maximum bytes" Compacted partition maximum bytes: 1188716932
If you detect a table with a huge partition there is a way to drill down even further starting from scylla 2.3 using large partition table