openzipkin/zipkin

Zipkin is a distributed tracing system

Adrian Cole

@adriancole

hi. not sure your networking setup, but you could decide to use the docker HEALTHCHECK instead

ex here's the definition from the image

# We use start period of 30s to avoid marking the container unhealthy on slow or contended CI hosts.
#
# If in production, you have a 30s startup, please report to https://gitter.im/openzipkin/zipkin
# including the values of the /health and /info endpoints as this would be unexpected.
HEALTHCHECK --interval=5s --start-period=30s --timeout=5s CMD ["docker-healthcheck"]

Halil Onur Çeçen

@onurccn

Hi all, just joined to the tracing world and already i have a question.
So we have just setup 2 zipkin servers with 5 node cassandra and 3 node kafka. We can receive/store traces without issues. But when i try to search traces with min or max duration parameter set, server gives

Caused by: com.datastax.driver.core.exceptions.NoHostAvailableException: All host(s) tried for query failed (tried: /10.110.220.8:9042 (com.datastax.driver.core.exceptions.BusyPoolException: [/10.110.220.8:9042] Pool is busy (no available co
nnection and the queue has reached its max size 40960)), /10.110.220.9:9042 (com.datastax.driver.core.exceptions.BusyPoolException: [/10.110.220.9:9042] Pool is busy (no available connection and the queue has reached its max size 40960)), /10.110.220.6:9042 (com.datastax.
driver.core.exceptions.BusyPoolException: [/10.110.220.6:9042] Pool is busy (no available connection and the queue has reached its max size 40960)), /10.110.220.7:9042, /10.110.220.5:9042 [only showing errors of first 3 hosts, use getErrors() for more details])

errors. Im not sure where to look for debugging so i am open to suggestions.
Also we currently cant calculate dependencies because cron jobs not finishing before the next one starts. So this issue might be related to performance of our cassandra cluster.
cassandra: 3.11.8,
zipkin: 2.21
Thanks,

34 replies

Vikas Phonsa

@vikasphonsa

Newbie to tracing. When/Why would you need an opencensus exporter to send spans to zipkin ? looking at https://zipkin.io/pages/architecture.html, couldn't figure out what role an exporter plays. seems like traces can go directly to zipkin collector ?

2 replies

Adrian Cole

@adriancole

Our LINE friends are on!! https://linedevday.linecorp.com/2020/en/live/Day2/casual

Marcin Grzejszczak

@marcingrzejszczak

hi! I can't find changelog / release notes for Brave - where can I find those?

2 replies

xuhao-xh

@xuhao-xh

i use this "STORAGE_TYPE=elasticsearch ES_HOSTS=xxxx ES_USERNAME=xxx ES_PASSWORD=xxxx java -jar zipkin-server-2.22.2-exec.jar --logging.level.zipkin2=INFO" to start my zipkin but i failed,why?please help me

1 reply

Adrian Cole

@adriancole

in case someone wants to watch https://linedevday.linecorp.com/2020/en/live/Day3/DPP_OSS

Scott Greenup

@scottgreenup

Hello everyone! :) I've just started instrumenting some services at work and hit a snag. I'm trying to work out what's best practice when creating a new span. The wording in https://zipkin.io/pages/instrumenting.html is a little ambiguous to me. Should I be creating a new span before I make a request (and send the new spanId with that request) or should the server receiving that request start a new span? I gather the choice is also up to me, but I'm wondering what the common pattern is.

Thanks in advance!

3 replies

Scott Greenup

@scottgreenup

Just a thought, is there a good resource to read about best practices around when to create new spans?

José Carlos Chávez

@jcchavezs

So in general if you add instrumentation to your libraries (e.g. http client, http server, etc) you don't need to create a span but they will be created for you.

Scott Greenup

@scottgreenup

That's the case for most of our stuff. Sadly we have one legacy library that isn't instrumented, so I'm having to wrap it.

I'm wondering if I pass the current span information along, or create a new one before sending it along.

But that's good advice in general :)

Adrian Cole

@adriancole

https://github.com/openzipkin/brave/releases/tag/5.13.2

3 replies

milind-techno

@milind-techno

Hello, Any recommendations for zipkin agents? I mean , agents which can sit alongside my application and send data to zipkin servers? Any specific recommendations please.
( Btw - I had a look at web and it seems skywalking agent is one , are there are alternatives especially from open-zipkin itself?

1 reply

Ucky.Kit

@uckyk

19 replies

i run zipkin-ui/Dockerfile in k8s,and how to deal with this error , its show me operate timeout .i dont know what happened...

Ucky.Kit

@uckyk

It is run by DockerFile in zipkin, without any modification, and the build is unsuccessful several times

Adrian Cole

@adriancole

the trick is this.. https://github.com/openzipkin/zipkin/blob/master/.github/workflows/readme_test.yml#L106-L107

if you are externally building with maven, then sharing with docker, you have to allow it

joybestourous

@joybestourous

What makes a span render red? Currently testing something and seeing no error annotations (apart from one error = false), but the span is red

Bas van Beek

@basvanbeek

Because if there is no error, there should not be an error tag at all

joybestourous

@joybestourous

thank you!

Bas van Beek

@basvanbeek

We treat error tag as the error and the value is for you to signal cause

José Carlos Chávez

@jcchavezs

What instrumentation are you using @joybestourous ?

joybestourous

@joybestourous

zipkin lens

1 reply

Tugay Ersoy

@Admiralkheir

hi, guys i have question about persistance. How can I configure zipkin(using for distributed tracing) with persisted. I don't want to lose trace logs after zipkin pod restarted

Bas van Beek

@basvanbeek

most common scenario is setting up an Elasticsearch cluster and using that as the storage backend

ES has a pretty decent operator so install should not be too hard

and you can instruct Zipkin to use ES by env. vars

https://github.com/openzipkin/zipkin/tree/master/zipkin-server#elasticsearch-storage

chuwuu

@chuwuu

hi，guys，how to configure zipkin to trace the grpc server, consumer can't capture the grpc requests according to official guide

Steve Conover

@sconover

Hi everyone, I've been googling around for a set of samples/example traces to do some analysis / perf testing with. Does anyone know of a zipkin sample trace dataset ... or any tracing dataset of this nature? (Ideally this dataset would be relatively large.)

Tugay Ersoy

@Admiralkheir

@basvanbeek i did like you said but now i cant see dependencies page

it is empty but traces are coming

Tugay Ersoy

@Admiralkheir

using elast 7.10.0

Bas van Beek

@basvanbeek

@Admiralkheir that is correct. you need to run a separate job when using external storage to aggregate the dependency links. See: https://github.com/openzipkin/zipkin-dependencies

@chuwuu there is various instrumentation for gRPC clients and servers. It would help to know what you are trying to instrument (which language/framework/etc)

Tugay Ersoy

@Admiralkheir

got it,if i dont use this job for 2 days. After that run this job is it linking all of them? or only this day?

Bas van Beek

@basvanbeek

as described in the repo, if you run it just before midnight you will have a topology view from the previous day. Each day needs a new one

Ben Iofel

@benwaffle

If i have two trees of spans in one file, with two SERVER spans whose parentId is not in the tree, Zipkin only shows one of them. That's not a bug in zipkin, right? It's a bug in the tracing library?

https://gist.github.com/benwaffle/30d64099f94dae70a02d9c687199d525

Bas van Beek

@basvanbeek

So it seems you're missing the root span then. so you have 2 fragments of the same trace where the originating span is missing.

Ben Iofel

@benwaffle

oh interesting, does the root span look different? or do you know just because I have 2 trees?

Where communities thrive

People

Repo info

Activity