Add process metadataedit

The add_process_metadata processor enriches events with information from running processes, identified by their process ID (PID).

  - add_process_metadata:
      match_pids: [system.process.ppid]
      target: system.process.parent

The fields added to the event look as follows:

"process": {
  "name":  "systemd",
  "title": "/usr/lib/systemd/systemd --switched-root --system --deserialize 22",
  "exe":   "/usr/lib/systemd/systemd",
  "args":  ["/usr/lib/systemd/systemd", "--switched-root", "--system", "--deserialize", "22"],
  "pid":   1,
  "ppid":  0,
  "start_time": "2018-08-22T08:44:50.684Z",
"container": {
  "id": "b5285682fba7449c86452b89a800609440ecc88a7ba5f2d38bedfb85409b30b1"

Optionally, the process environment can be included, too:

  "env": {
    "HOME":       "/",
    "TERM":       "linux",
    "BOOT_IMAGE": "/boot/vmlinuz-4.11.8-300.fc26.x86_64",
    "LANG":       "en_US.UTF-8",

It has the following settings:

List of fields to lookup for a PID. The processor will search the list sequentially until the field is found in the current event, and the PID lookup will be applied to the value of this field.
(Optional) Destination prefix where the process object will be created. The default is the event’s root.
(Optional) List of fields to add. By default, the processor will add all the available fields except process.env.
(Optional) When set to false, events that don’t contain any of the fields in match_pids will be discarded and an error will be generated. By default, this condition is ignored.
(Optional) By default, if a target field already exists, it will not be overwritten and an error will be logged. If overwrite_keys is set to true, this condition will be ignored.
(Optional) By default, the process.env field is not output, to avoid leaking sensitive data. If restricted_fields is true, the field will be present in the output.
(Optional) By default, the host_path field is set to the root directory of the host /. This is the path where /proc is mounted. For different runtime configurations of Kubernetes or Docker, the host_path can be set to overwrite the default.
(Optional) By default, the cgroup_prefixes field is set to /kubepods and /docker. This is the prefix where the container ID is inside cgroup. For different runtime configurations of Kubernetes or Docker, the cgroup_prefixes can be set to overwrite the defaults.

(Optional) By default, the container id is extracted from cgroup file based on cgroup_prefixes. This can be overwritten by specifying regular expression with capture group for capturing container id from cgroup path. Examples:

  1. ^\/.+\/.+\/.+\/([0-9a-f]{64}).* will match the container id of a cgroup like /kubepods/besteffort/pod665fb997-575b-11ea-bfce-080027421ddf/b5285682fba7449c86452b89a800609440ecc88a7ba5f2d38bedfb85409b30b1
  2. ^\/.+\/.+\/.+\/docker-([0-9a-f]{64}).scope will match the container id of a cgroup like /kubepods.slice/kubepods-burstable.slice/kubepods-burstable-pod69349abe_d645_11ea_9c4c_08002709c05c.slice/docker-80d85a3a585f1575028ebe468d83093c301eda20d37d1671ff2a0be50fc0e460.scope
  3. ^\/.+\/.+\/.+\/crio-([0-9a-f]{64}).scope will match the container id of a cgroup like /kubepods.slice/kubepods-burstable.slice/kubepods-burstable-pod69349abe_d645_11ea_9c4c_08002709c05c.slice/crio-80d85a3a585f1575028ebe468d83093c301eda20d37d1671ff2a0be50fc0e460.scope
(Optional) By default, the cgroup_cache_expire_time is set to 30 seconds. This is the length of time before cgroup cache elements expire in seconds. It can be set to 0 to disable the cgroup cache. In some container runtimes technology like runc, the container’s process is also process in the host kernel, and will be affected by PID rollover/reuse. The expire time needs to set smaller than the PIDs wrap around time to avoid wrong container id.