FluentD to SigNoz
If you use fluentD to collect logs in your stack with this tutotrial you will be able to send logs from fluentD to SigNoz.
At SigNoz we use opentelemetry collector to recieve logs which supports the fluentforward protocol. So you can forward your logs from your fluentD agent to opentelemetry collector.
Collect Logs Using FluentD in SigNoz cloud
Add otel collector binary to your VM by following this guide.
Add fluentforward reciever to your
config.yaml
receivers: fluentforward: endpoint: 0.0.0.0:24224
Here we have used port 24224 for listening in fluentforward protocol, but you can change it to a port you want. You can read more about fluentforward receiver here.
Modify your
config.yaml
and add the above receiverservice: .... logs: receivers: [otlp, fluentforward] processors: [batch] exporters: [otlp]
Add the following to your fluentD config to forward the logs to otel collector.
<match <directive>> @type forward send_timeout 60s recover_wait 10s hard_timeout 60s <server> name myserver1 host localhost port 24224 </server> </match>
In this config we are matching a directive and forwarding logs to the otel collector which is listening on port 24224. Replace
<directive>
with your directive name. Also we are assuming that you are running the fluentD binary on the host. If not, the value ofhost
might change depending on your environment.Once you make this changes you can restart fluentD and otel-binary, and you will be able to see the logs in SigNoz.
To properly transform your existing log model into opentelemetry log model you can use the different processors provided by opentelemetry. link
eg:-
processors:
logstransform:
operators:
- type: trace_parser
trace_id:
parse_from: attributes.trace_id
span_id:
parse_from: attributes.span_id
- type: remove
field: attributes.trace_id
- type: remove
field: attributes.span_id
The operations in the above processor will parse the trace_id and span_id from log to opentelemetry log model and remove them from attributes.
Collect Logs Using FluentD in Self-Hosted SigNoz
Steps to recieve logs from FluentD:
Add fluentforward reciever to your
otel-collector-config.yaml
which is present insidedeploy/docker/clickhouse-setup
receivers: fluentforward: endpoint: 0.0.0.0:24224
Here we have used port 24224 for listening in fluentforward protocol, but you can change it to a port you want. You can read more about fluentforward receiver here.
Uncomment the exporter and pipleline for logs and make the following change in
otel-collector-config.yaml
exporters: ... clickhouselogsexporter: dsn: tcp://clickhouse:9000/ timeout: 5s sending_queue: queue_size: 100 retry_on_failure: enabled: true initial_interval: 5s max_interval: 30s max_elapsed_time: 300s ... service: ... logs: receivers: [ otlp, fluentforward ] processors: [ batch ] exporters: [ clickhouselogsexporter ]
Here we are adding our clickhouse exporter and creating a pipeline which will collect logs from
fluentforward
receiver, processing it using batch processor and export it to clickhouse.Expose the port in port for otel-collector in
docker-compose.yaml
file present indeploy/docker/clickhouse-setup
otel-collector: ... ports: - "24224:24224"
Change the fluentD config to forward the logs to otel collector.
<source> @type sample sample [{"message": "my log data", "source": "myhost"}, {"message": "my log data 1", "source": "myhost1"}] tag sample rate 10000 </source> <match sample> @type forward send_timeout 60s recover_wait 10s hard_timeout 60s <server> name myserver1 host <otel-collector-host> port 24224 </server> </match>
In this example we are generating sample logs and then forwarding them to the otel collector which is listening on port 24224.
<otel-collector-host>
has to be replaced by the host where otel-collector is running. For more info check troubleshooting.Once you make this changes you can restart fluentD and SignNoz, and you will be able to see the logs in SigNoz.
To properly transform your existing log model into opentelemetry log model you can use the different processors provided by opentelemetry. link
eg:-
processors: logstransform: operators: - type: trace_parser trace_id: parse_from: attributes.trace_id span_id: parse_from: attributes.span_id - type: remove field: attributes.trace_id - type: remove field: attributes.span_id
The operations in the above processor will parse the trace_id and span_id from log to opentelemetry log model and remove them from attributes.