Skip to content
This repository has been archived by the owner on May 18, 2020. It is now read-only.

all-in-one is extremely slow #135

Open
talr1 opened this issue May 12, 2020 · 0 comments
Open

all-in-one is extremely slow #135

talr1 opened this issue May 12, 2020 · 0 comments

Comments

@talr1
Copy link

talr1 commented May 12, 2020

Requirement - what kind of business use case are you trying to solve?

work with jaeger locally and in Kubernetes

Problem - what in Jaeger blocks you from solving the requirement?

using the all-in-one deployment locally in docker-compose and also in k8s cluster is extremely slow.
i have 5 NodeJS micro-services instrumented with opentracing and jaeger reporting spans to jaeger all-in-one. a simple http request which results in ~100 spans takes almost 90 sec. when tracing is turned off (by using no-op tracer), the request ends in 1-2 secs.
i assume the instrumentation code is not to blame but the actual spans processing and sending to jaeger backend.

looking on traces, i see that outgoing requests from any micro services seems to take quite long, ~around 10-15 sec with no explanation. using a UDP port (6832) AFAIK should not block or cause such hangs.

i don't find any instructions how to debug the issue.
I've tried to add grafana and Prometheus to my compose to metrics but i an;'t find a proper dashboard, nor i don't see any metrics scraping happening. (I've added a task for all-in-one:14269 as described here: https://www.jaegertracing.io/docs/1.17/monitoring

it would be great to have a complete example that includes all-in-one Jaeger + grafana + (with pre-configured dashboard) Prometheus to allow easy Jaeger monitoring/debugging.

AFAIK, Jaeger is built for huge spans processing, and i can't see why all-in-one can't process ~100 spans for a single request (no other requests happens, besides a few /health traces, but i assume they are negligible)

local env:

OS: Windows_NT x64 10.0.18363
Node.js: 12.16.1
"jaeger-client": "^3.18.0",
"opentracing": "^0.14.4",
jaeger all-in-one (latest, i assume 1.17)
docker-compose version 1.25.4, build 8d51620a
Docker version 19.03.8, build afacb8b

slowness happens also in our k8s cluster where i've deployed the all-in-one image

can't use tracing as it extremely slow

Proposal - what do you suggest to solve the problem or improve the existing situation?

Any open questions to address

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant