Loki: Increase (query/search) performance for the Docker / single binary?

[frpet1]

Hi,

Then running the “docker-compose” version of Loki, all queries are running on a single CPU-core, which is inefficient especially then you have multiple cores available (and go is really good at doing parallel tasks) .

I found this article, but it’s for running loki in k8s.

How to improve Loki Performance

Hello Loki Team! I would like some help tweaking the performance of our Loki setup. I’ve read blog posts about blazing speeds but can’t seem to achieve this. The read path is what I am trying to optimize. The current setup is Loki 2.0 in microservices mode deployed to AWS using S3 and DynamoDB for chunks/index. We are using memberlist for the Ring. ElasticCache Redis has been added as the cache. We are running 2 query frontends, 3 distributors, 3 ingesters, 6 queriers and 1 table manager co…

And I guess most bigger installations has moved to k8s already.

I tried to add some configs from the Slack channel, but that had no effect on parallelism AFAICT. But from the post above, I interpret that there need to be more frontend_workers too?

query_range:
  align_queries_with_step: true
  max_retries: 10
  split_queries_by_interval: 15m
  split_queries_by_day: true
  parallelise_shardable_queries: true
  cache_results: true
  results_cache:
    cache:
      redis:
        endpoint: loki:6379
        expiration: 86400s
        db: 1
        timeout: 60s

If possible, How would a single binary config for more parallel “search workers” look like?

[frpet1]

To illustrate this, this happens when I perform a query in Loki/Grafana then I do a query+filter like {label=somelabel} |= “something_I_search_for”

(second row shows the thread going 100% while the others have basically no load)

[dujas]

Hi, did you solve the issue yet? I am running loki in local build mode by following the doker compose design as well.