About threshold best practice for fdb metrics

Using FoundationDB
1

Hi, Experts

I am configuring fdb metrics alert threshold as following, I am not sure if they are reasonable? Any comment? Or any best practice for these thresholds? And I also wonder if there is any fdb performance tuning guide for reference? Thanks!

[Database CPU]
CPU_USAGE_CORES_WARNING=0.5
CPU_USAGE_CORES_CRITICAL=0.7

[Database Memory]
MEMORY_AVAILABLE_BYTES_CRITICAL=100000000
MEMORY_AVAILABLE_BYTES_WARNING=200000000

[Database Disk]
DISK_BUSY_WARNING=0.5
DISK_BUSY_CRITICAL=0.7
DISK_FREE_BYTES_WARNING=100000000
DISK_FREE_BYTES_CRITICAL=50000000
WORST_QUEUE_BYTES_STORAGE_SERVER_WARNING=100000
WORST_QUEUE_BYTES_LOG_SERVER_WARNING=100000
KVSTORE_FREE_BYTES_WARNING=10000000
KVSTORE_FREE_BYTES_CRITICAL=5000000

[Database Network]
NETWORK_CONNECTION_ERRORS_PER_SECOND_CRITICAL=10
NETWORK_CONNECTION_ERRORS_PER_SECOND_WARNING=5

[Database Latency]
TRANSACTION_START_SECONDS_CRITICAL=0.05
TRANSACTION_START_SECONDS_WARNING=0.01
COMMIT_SECONDS_CRITICAL=0.05
COMMIT_SECONDS_WARNING=0.01
READ_SECONDS_CRITICAL=0.05
READ_SECONDS_WARNING=0.01

And about following 2 metrics what threshold is reasonable?

  1. Worst Data Lag Storage Server Seconds
  2. Storage Data Lag Seconds
2

Here is a list of monitored metrics:

Admittedly, the doc is a bit old and may miss new metrics or need a little update.