Workload management
Workload managers control when additional containers are started or idle containers are stopped. See [root:Netplex_intro].workload for additional documentation how they are configured.
A constant number of threads is created (the int argument).
restart
: If threads
crash, new threads are created until the specified number is again
reached. This is on by default.
greedy_accepts
: whether greedy accepts are allowed (default: false)
max_jobs_per_thread
: if passed, limits the number of jobs (connections)
that can be simultaneously processed by a thread/process. By default
there is no limit.
Reads a workload_manager section like
workload_manager {
type = "constant";
threads = <n>;
max_jobs_per_thread = <n>;
greedy_accepts = <bool>;
}
How many jobs every thread can execute concurrently until it is considered as fully loaded. For configurations where the threads can only handle one connection at a time this number must be 1.
The number of jobs every thread can execute with normal
service quality. Must be less than or equal to
max_jobs_per_thread
The manager starts as many threads as required to ensure that this number of jobs can be executed. Must be at least 1.
If more job capacity is available than this number, threads are
terminated. Must be greater than or equal to
min_free_job_capacity
.
After this number of seconds a free thread can be terminated
The manager does not start more threads than this number
Whether greedy accepts are permitted
Reads a workload_manager section like
workload_manager {
type = "dynamic";
max_jobs_per_thread = <n>;
min_free_jobs_capacity = <n>;
max_free_jobs_capacity = <n>;
max_threads = <n>;
greedy_accepts = <bool>;
}
All built-in workload manager factories