# README
Kevent
is the fundamental data structure for transporting kernel events.
Each kernel event structure contains a series of canonical fields that describe the nature of the event such as its name, the process identifier that generated the event and such. The following is the list of all canonical fields.
- Sequence is a monotonically increasing integer value that uniquely identifies an event. The sequence value is guaranteed to increment monotonically as long as the machine is not rebooted. After the restart, the sequence is restored to the zero value.
- PID represents the process identifier that triggered the kernel event.
- TID is the thread identifier connected to the kernel event.
- CPU designates the logical CPU core on which the event was originated.
- Name is the human-readable event name such as
CreateProcess
orRegOpenKey
. - Timestamp denotes the timestamp expressed in nanosecond precision as the instant the event occurred.
- Category designates the category to which the event pertains, e.g.
file
orthread
. Each particular category is explained thoroughly in the next sections. - Description is a short explanation about the purpose of the event. For example,
CreateFile
kernel event creates or opens a file, directory, I/O device, pipe, console buffer or other block/pseudo device. - Host represents the host name where the event was produced.
Parameters
Also called as kparams
in Fibratus parlance, contain each of the event's parameters. Internally, they are modeled as a collection of key/value pairs where the key is mapped to the structure consisting of parameter name, parameter type and the value. An example of the parameter tuple could be the dip
parameter
that denotes a destination IP address with value 172.17.0.2
and therefore IPv4
type. Additionally, parameter types can be scalar values, strings, slices, enumerations, and timestamps among others.
Process state
Each event stores the process state that represents an extended information about the process including its allocated resources such as handles, dynamically-linked libraries, exported environment variables and other attributes. The process state internals are thoroughly explained in the Process events section.
Metadata
Metadata are an arbitrary sequence of tags in form of key/value pairs that you can squash into the event on behalf of transformers. A tag can be virtually any string data that you find meaningful to either identify the event or apply filtering/grouping once event is persisted in the data store.