We are onboarding 17K devices/applications via API. API response slowed down significantly overnight,
the CPU is high on the appliance targeted by the api caller.
Appliance with high CPU.
Other Appliance
Release : 4.1
Component :
Mysql and proxysql services were running a bit high on utilization but not to the point of failure (140% and 60% respectfully). The server was defined using 8 cpu and 32 gig of ram.
The performance for the API has no user configurable customizations but we do pre-allocate a % of the ram to Java services and mysql at the time the appliance starts and this is calculated based on the amount of ram present. Even though the ram usage remained constant at 57% we suggest raising the amount of ram to 64 gig which will preallocate more ram for these processes which will improve performance. Networking issues can also play a part in this as all records being added or modified must be replicated throughout the cluster.