3

I am trying to lookup a key from a record and use it as logstash prefix in fluent bit. But that's not happening and Logstash_Prefix is not being replaced by Logstash_Prefix_Key even though the specified key exists in the enriched log from kubernetes filter.

The ideal behaviour of a kubernetes filter is to enrich the logs read from input path via input plugin with kubernetes data such as pod name, pod id, namespace name etc. And when the logs after applying the filter is pushed to output source via es output plugin. I used Logstash_Prefix_Key to get the key kubernetes.pod_name and gave Logstash_Prefix as pod_name. Even though I'm able to see kubernetes.pod_name key in Kibana, the logs are getting stored in the prefix pod_name (which means Logstash_Prefix_Key is not found tn log records so it uses Logstash_Prefix).

Code sample

input-kubernetes.conf: |
    [INPUT]
        Name              tail
        Tag               kube.*
        Path              /var/log/containers/*.log
        Parser            docker
        DB                /var/log/flb_kube.db
        Mem_Buf_Limit     2GB
        Skip_Long_Lines   On
        Refresh_Interval  10

  filter-kubernetes.conf: |
    [FILTER]
        Name                kubernetes
        Match               kube.*
        Kube_URL            https://kubernetes.default.svc.cluster.local:443
        Merge_Log           Off
        K8S-Logging.Parser  On

  output-elasticsearch.conf: |
    [OUTPUT]
        Name            es
        Match           kube.*
        Host            ${FLUENT_ELASTICSEARCH_HOST}
        Port            ${FLUENT_ELASTICSEARCH_PORT}
        HTTP_User       ${FLUENT_ELASTICSEARCH_USER}
        HTTP_Passwd     ${FLUENT_ELASTICSEARCH_PASSWORD}
        Logstash_Format On
        Logstash_Prefix pod_name
        Logstash_Prefix_Key kubernetes.pod_name
        Retry_Limit     False

Since I am new to EFK stack, Could someone help me with this

Nick
  • 1,882
  • 11
  • 16
coder here
  • 179
  • 3
  • 10

3 Answers3

3

UPD: it's now supported! https://github.com/fluent/fluent-bit/issues/421#issuecomment-766912018 Should be in Fluent Bit v1.7 release!

Dynamic ElasticSearch indexes are not supported in FluentBit at the moment. Here's a related issue: https://github.com/fluent/fluent-bit/issues/421. You can only specify a string (hardcoded) prefixes.

The workaround is to use a fluentd log collector instead, which supports dynamic indexes: https://docs.fluentd.org/output/elasticsearch#index_name-optional. There's a community chart for it: https://github.com/helm/charts/tree/master/stable/fluentd

UPD: it's now supported! https://github.com/fluent/fluent-bit/issues/421#issuecomment-766912018 Should be in Fluent Bit v1.7 release!

Max Lobur
  • 5,662
  • 22
  • 35
2

Was trying to do the same recently and though what Max Lobur said above is true about fluentbit not having support for this prior to the not yet released version 1.7. However, I was still able to achieve this with the current version using the nest filter, see https://docs.fluentbit.io/manual/pipeline/outputs/elasticsearch .under the Logstash_Prefix_Key it says

When included: the value in the record that belongs to the key will be looked up and over-write the Logstash_Prefix for index generation. If the key/value is not found in the record then the Logstash_Prefix option will act as a fallback. Nested keys are not supported (if desired, you can use the nest filter plugin to remove nesting)

The last sentence is about nested keys not supported however you can still use them if you use the nest filter to lift them up a level.

In your case the pod_name is nested under kubernetes, to still be able to use it, you would have to lift it out of that level. see nest example here.

Here's how to make it work in your case:

filter-kubernetes.conf: |
[FILTER]
    Name                kubernetes
    Match               kube.*
    Kube_URL            https://kubernetes.default.svc.cluster.local:443
    Merge_Log           Off
    K8S-Logging.Parser  On

[FILTER]
    Name                nest
    Match               *
    Operation           lift
    Nested_under        kubernetes
    Add_prefix          kubernetes_

output-elasticsearch.conf: |
[OUTPUT]
    Name            es
    Match           kube.*
    Host            ${FLUENT_ELASTICSEARCH_HOST}
    Port            ${FLUENT_ELASTICSEARCH_PORT}
    HTTP_User       ${FLUENT_ELASTICSEARCH_USER}
    HTTP_Passwd     ${FLUENT_ELASTICSEARCH_PASSWORD}
    Logstash_Format On
    Logstash_Prefix pod_name
    Logstash_Prefix_Key kubernetes_pod_name
    Retry_Limit     False

what we are doing here is lifting everything inside the kubernetes object up a level and prefixing them with kubernetes_, so your pod_name will become kubernetes_pod_name. You then pass the kubernetes_pod_name to Logstash_prefix_key. The value of kubernetes_pod_name is then use for index generation and would only resort back to logstash_prefix if no key/value pair exist for the kubernetes_pod_name

Ola
  • 161
  • 4
1

You can use:

Logstash_Prefix_Key kubernetes['pod_name']

This is working on my machine using the docker image: fluent/fluent-bit:1.7

mrk
  • 640
  • 8
  • 16