Advanced Statistics

The following advanced statistics require operating system support to collect:

  • cpu_wait_time_ms

  • disk_time_ms

  • disk_physical_read_b

  • disk_physical_write_b

  • memory_major_faults

If SingleStore DB failed to collect an advanced statistic, the corresponding column in workload profiling views will contain NULL values and warnings will be returned by queries against the views.

In particular, cpu_wait_time_ms requires a Linux kernel compiled with CONFIG_SCHEDSTAT enabled. This flag is enabled by default on all Ubuntu and Red Hat Linux installations, but is omitted from certain versions of Debian Linux.


Advanced statistics are not collected by default in MemSQL 5.8. This is due to performance overhead of up to 10-40µS per query, equivalent to 5-10% overhead for the simple query select 1. We strongly recommend enabling advanced statistics, as they are critical to detecting disk, CPU, and memory saturation.

You may enable collection by setting the global variable read_advanced_counters to on on all nodes in the cluster. For more information on setting global variables, see the engine variables overview.

In order to generate these statistics, SingleStore reads: /proc/[tid]/stat//proc/[pid]/io/proc/[pid]/schedstat, and /proc/[pid]/sched -- if any of those are missing, some columns in the information schema tables may be null, even with @@read_advanced_counters set to on.