Flow log data types

Big picture

Calico Enterprise sends the following data to Elasticsearch.

The following table details the key/value pairs in the JSON blob, including their Elasticsearch datatype.

Name	Datatype	Description
host	keyword	Name of the node that collected the flow log entry.
start_time	date	Start time of log collection in UNIX timestamp format.
end_time	date	End time of log collection in UNIX timestamp format.
action	keyword	- allow: Calico Enterprise accepted the flow. - deny: Calico Enterprise denied the flow.
bytes_in	long	Number of incoming bytes since the last export.
bytes_out	long	Number of outgoing bytes since the last export.
dest_ip	ip	IP address of the destination pod. A null value indicates aggregation.
dest_name	keyword	Contains one of the following values: - Name of the destination pod. - Name of the pod that was aggregated or the endpoint is not a pod. Check dest_name_aggr for more information, such as the name of the pod if it was aggregated.
dest_name_aggr	keyword	Contains one of the following values: - Aggregated name of the destination pod. - pvt: endpoint is not a pod. Its IP address belongs to a private subnet. - pub: endpoint is not a pod. Its IP address does not belong to a private subnet. It is probably an endpoint on the public internet.
dest_namespace	keyword	Namespace of the destination endpoint. A - means the endpoint is not namespaced.
dest_port	long	Destination port. Not applicable for ICMP packets.
dest_service_name	keyword	Name of the destination service. A - means the original destination did not correspond to a known Kubernetes service (e.g. a services ClusterIP).
dest_service_namespace	keyword	Namespace of the destination service. A - means the original destination did not correspond to a known Kubernetes service (e.g. a services ClusterIP).
dest_service_port	keyword	Port name of the destination service. A - means : - the original destination did not correspond to a known Kubernetes service (e.g. a services ClusterIP), or - the destination port is aggregated. A * means there are multiple service port names matching the destination port number.
dest_type	keyword	Destination endpoint type. Possible values: - wep: A workload endpoint, a pod in Kubernetes. - ns: A Networkset. If multiple Networksets match, then the one with the longest prefix match is chosen. - net: A Network. The IP address did not fall into a known endpoint type.
dest_labels	array of keywords	Labels applied to the destination pod. A hyphen indicates aggregation.
dest_domains	array of keywords	Find all the destination domain names for use in a DNS policy by examining dest_domains. The field displays information on the top-level domains linked to the destination IP. Applies to flows reported from the source to destinations outside the cluster. If flowLogsDestDomainsByClient is disabled, having dest_domains: ["A"] doesn't guarantee that the flow corresponds to a connection with domain name A. The destination IP may also be linked to other domain names not yet captured by Calico.
reporter	keyword	- src: flow came from the pod that initiated the connection. - dst: flow came from the pod that received the initial connection.
num_flows	long	Number of flows aggregated into this entry during this export interval.
num_flows_completed	long	Number of flows that were completed during the export interval.
num_flows_started	long	Number of flows that were started during the export interval.
num_process_names	long	Number of unique process names aggregated into this entry during this export interval.
num_process_ids	long	Number of unique process ids aggregated into this entry during this export interval.
num_process_args	long	Number of unique process args aggregated into this entry during this export interval.
nat_outgoing_ports	array of ints	List of NAT outgoing ports for the packets that were Source NAT'd in the flow
packets_in	long	Number of incoming packets since the last export.
packets_out	long	Number of outgoing packets since the last export.
proto	keyword	Protocol.
policies	array of keywords	List of policies that interacted with this flow. See Format of the policies field.
process_name	keyword	The name of the process that initiated or received the connection or connection request. This field will have the executable path if flowLogsCollectProcessPath is enabled. A "-" indicates that the process name is not logged. A "*" indicates that the per flow process limit has exceeded and the process names are now aggregated.
process_id	keyword	The process ID of the corresponding process (indicated by the process_name field) that initiated or received the connection or connection request. A "-" indicates that the process ID is not logged. A "*" indicates that there are more than one unique process IDs for the corresponding process name.
process_args	array of strings	The arguments with which the executable was invoked. The size of the list depends on the per flow process args limit.
source_ip	ip	IP address of the source pod. A null value indicates aggregation.
source_name	keyword	Contains one of the following values: - Name of the source pod. - Name of the pod that was aggregated or the endpoint is not a pod. Check source_name_aggr for more information, such as the name of the pod if it was aggregated.
source_name_aggr	keyword	Contains one of the following values: - Aggregated name of the source pod. - pvt: Endpoint is not a pod. Its IP address belongs to a private subnet. - pub: the endpoint is not a pod. Its IP address does not belong to a private subnet. It is probably an endpoint on the public internet.
source_namespace	keyword	Namespace of the source endpoint. A - means the endpoint is not namespaced.
source_port	long	Source port. A null value indicates aggregation.
source_type	keyword	The type of source endpoint. Possible values: - wep: A workload endpoint, a pod in Kubernetes. - ns: A Networkset. If multiple Networksets match, then the one with the longest prefix match is chosen. - net: A Network. The IP address did not fall into a known endpoint type.
source_labels	array of keywords	Labels applied to the source pod. A hyphen indicates aggregation.
original_source_ips	array of ips	List of external IP addresses collected from requests made to the cluster through an ingress resource. This field is only available if capturing external IP addresses is configured.
num_original_source_ips	long	Number of unique external IP addresses collected from requests made to the cluster through an ingress resource. This count includes the IP addresses included in the original_source_ips field. This field is only available if capturing external IP addresses is configured.
tcp_mean_send_congestion_window	long	Mean tcp send congestion window size. This field is only available if flowLogsEnableTcpStats is enabled
tcp_min_send_congestion_window	long	Minimum tcp send congestion window size. This field is only available if flowLogsEnableTcpStats is enabled
tcp_mean_smooth_rtt	long	Mean smooth RTT in micro seconds. This field is only available if flowLogsEnableTcpStats is enabled
tcp_max_smooth_rtt	long	Maximum smooth RTT in micro seconds. This field is only available if flowLogsEnableTcpStats is enabled
tcp_mean_min_rtt	long	Mean MinRTT in micro seconds. This field is only available if flowLogsEnableTcpStats is enabled
tcp_max_min_rtt	long	Maximum MinRTT in micro seconds. This field is only available if flowLogsEnableTcpStats is enabled
tcp_mean_mss	long	Mean TCP MSS. This field is only available if flowLogsEnableTcpStats is enabled
tcp_min_mss	long	Minimum TCP MSS. This field is only available if flowLogsEnableTcpStats is enabled
tcp_total_retransmissions	long	Total retransmitted packets. This field is only available if flowLogsEnableTcpStats is enabled
tcp_lost_packets	long	Total lost packets. This field is only available if flowLogsEnableTcpStats is enabled
tcp_unrecovered_to	long	Total unrecovered timeouts. This field is only available if flowLogsEnableTcpStats is enabled

Format of the policies field

The policies field contains a comma-delimited list of policy rules that matched the flow. Each entry in the list has the following format:

<index>|<tier name>|<policy name>|<action>|<rule index>

Where,

• <index> numbers the order in which the rules were hit, starting with 0.

  tip

  Sort the entries of the list by the <index> to see the order that rules were hit. The entries are displayed in random order due to the way they are stored in the datastore.

• <tier name> is the name of the policy tier containing the policy, or __PROFILE__ for a rule derived from a Profile resource (this is the internal datatype used to represent a Kubernetes namespace and its associated "default allow" rule).

• <policy name> is the name of the policy/profile; its format depends on the type of policy:

  • <tier>.<name> for Calico Enterprise GlobalNetworkPolicy.
  • <namespace>/knp.default.<name> for Kubernetes NetworkPolicy.
  • <namespace>/<tier>.<name> for Calico Enterprise NetworkPolicy.

  Staged policy names are prefixed with "staged:".

• <action> is the action performed by the rule; one of allow, deny, pass.

• <rule index> if non-negative, is the index of the rule that was matched within the policy, starting with 0. Otherwise, a special value:

  • -1 means the reporting endpoint was selected by the policy but no rule matched. The traffic hit the default action for the tier. In this case, the <policy name> is selected arbitrarily from the set of policies within the tier that apply to the endpoint.
  • -2 means "unknown". The rule index was not recorded.

Flow log example, with no aggregation

A flow log with aggregation level 0, no aggregation, might look like:

  {
    "start_time": 1597166083,
    "end_time": 1597166383,
    "source_ip": "192.168.47.9",
    "source_name": "access-6b687c8dcb-zn5s2",
    "source_name_aggr": "access-6b687c8dcb-*",
    "source_namespace": "policy-demo",
    "source_port": 42106,
    "source_type": "wep",
    "source_labels": {
      "labels": [
        "pod-template-hash=6b687c8dcb",
        "app=access"
      ]
    },
    "dest_ip": "192.168.138.79",
    "dest_name": "nginx-86c57db685-h6792",
    "dest_name_aggr": "nginx-86c57db685-*",
    "dest_namespace": "policy-demo",
    "dest_port": 80,
    "dest_type": "wep",
    "dest_labels": {
      "labels": [
        "pod-template-hash=86c57db685",
        "app=nginx"
      ]
    },
    "proto": "tcp",
    "action": "allow",
    "reporter": "dst",
    "policies": {
      "all_policies": [
        "0|default|policy-demo/default.access-nginx|allow"
      ]
    },
    "bytes_in": 388,
    "bytes_out": 1113,
    "num_flows": 1,
    "num_flows_started": 1,
    "num_flows_completed": 1,
    "packets_in": 6,
    "packets_out": 5,
    "http_requests_allowed_in": 0,
    "http_requests_denied_in": 0,
    "original_source_ips": null,
    "num_original_source_ips": 0,
    "host": "bz-n8kf-kadm-node-1",
    "@timestamp": 1597166383000
  }

The log shows an incoming connection reported by the destination node, allowed by a policy on port 80. The start_time and end_time describe the aggregation period (5 min.) During this interval, one flow ("num_flow": 1) was recorded. At higher aggregation levels, flows from endpoints performing the same operation and originating from the same Deployment/ReplicaSet are grouped into a single log. In this example, the common source endpoints that are prefixed with access-6b687c8dcb-. Parameters like source_ip may be dropped and set to null, depending on the aggregation level. As aggregation levels increase, more flows will be grouped together based on your data. For more details on aggregation levels, see configure flow log aggregation.