m dirty settings are default:
vm.dirty_background_bytes = 0
vm.dirty_ratio = 20
vm.dirty_bytes = 0
vm.dirty_writeback_centisecs = 500
vm.dirty_expire_centisecs = 3000
If this is a NUMA environment, what does this show?:
>
> numactl --hardware
>
> available: 1 nodes (0)
node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
node 0 size: 124999 MB
node 0 free: 740 MB
node distances:
node 0
0: 10
Carlo Cabanilla
Datadog
the postgres replica unable to close connections
during these incidents? And why were reads on the replica so slow at the
time? Are there locks that are locking out the readers when there's lots of
update churn?
Carlo Cabanilla
Datadog
