I decided to introduce frontend into the current cluster and I would like to configure frontend_address
in frontend_worker instead of downstream_url
in frontend per the discussion in https://community.grafana.com/t/how-to-improve-loki-performance/42800.
Besides, I would like to put the 2 frontend services into nginx to make them load balancing, and here is the configuration of loki:
frontend:
log_queries_longer_than: 5s
#downstream_url: http://centos8:3100
compress_responses: true
frontend_worker:
frontend_address: centos8:9095
parallelism: 12
centos8 is the server that nginx locates in and its config file(based on https://github.com/grafana/loki/blob/main/production/docker/config/nginx-loki-gateway.conf) is like:
upstream query-frontend {
server centos8-3:9095;
server centos8-4:9095;
}
server {
listen 9095 http2;
location / {
grpc_pass grpc://query-frontend;
}
But once loki is spun up, error like below posted:
Aug 24 01:24:34 centos8-2 loki-linux-amd64[6725]: level=error ts=2021-08-23T17:24:34.719783648Z caller=frontend_processor.go:70 msg="error processing requests" address=192.168.209.133:9095 err="rpc error: code = Internal desc = stream terminated by RST_STREAM with error code: PROTOCOL_ERROR"
I think there must be some mis-configuration in nginx, please guide me how to move on.