Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • 08:15
    broucetrong edited #100
  • 08:14
    broucetrong opened #100
  • 07:42

    anuraaga on master

    Remove unused scss definitions … (compare)

  • 07:42
    anuraaga closed #2983
  • 07:33
  • 07:33
    Travis openzipkin/brave@7376780 (request.response) errored (3838)
  • 07:32
    Travis anuraaga/zipkin@6e43f75 (scss-to-css) passed (340)
  • 07:30
    adriancole commented #1092
  • 07:29
    adriancole review_requested #1092
  • 07:29
    adriancole opened #1092
  • 07:29

    adriancole on request.response

    WIP: Adds HttpResponse.request(… (compare)

  • 07:06
    anuraaga edited #2983
  • 07:05
    anuraaga review_requested #2983
  • 07:05
    anuraaga review_requested #2983
  • 07:05
    anuraaga opened #2983
  • 05:11

    adriancole on method-mandatory

    (compare)

  • 05:11

    adriancole on master

    Makes HTTP request method manda… (compare)

  • 05:11
    adriancole closed #1091
  • 05:11
    adriancole closed #1089
  • 05:11
    adriancole commented #1091
lusche
@lusche
thanks for the support! It was very helpful.
Adrian Cole
@adriancole
we never advocate custom servers because they are very expensive to support usually people mess up classpath and it takes hours or days, sometimes for a feature that works by default
so yeah I would recommend focus on the property reading aspect instead. good luck!
lusche
@lusche
make sens and should work.
Anders Clausen
@AndersClausen

Hi @adriancole When you say 'custom servers' does that include a spring boot service that uses these two dependencies?

        <dependency>
            <groupId>io.zipkin.java</groupId>
            <artifactId>zipkin-server</artifactId>
            <version>2.12.9</version>
        </dependency>
        <dependency>
            <groupId>io.zipkin.java</groupId>
            <artifactId>zipkin-autoconfigure-ui</artifactId>
            <version>2.12.9</version>
            <scope>runtime</scope>
        </dependency>

I just want to make sure I understand your statement correctly :-)

Adrian Cole
@adriancole
yep. folks who do this are on their own for support
Anders Clausen
@AndersClausen
@adriancole Thanks for clarifying that :-)
Adrian Cole
@adriancole
no problem
Javi
@Javigomez_gitlab
Hi all! I am very pleased to join this community. My name is Javi, and we are facing a issue in the NodeJS instrumentation. Specifically this issue: https://github.com/openzipkin/zipkin-js/issues/365#issuecomment-575665142. Adrian just suggested us to "if the data we have is related to reporting of unfinished spans, as it seems also explainable by a leaked context". Could anyone teach us how to debug if we are reporting this "unfinished spans"? How can we store the recorder info in a log file to study it? Thanks very much
Adrian Cole
@adriancole
hi @Javigomez_gitlab to test if unfinished spans are the problem, you can look at the json
if a span was unfinished, there would be no duration for it
so you can confirm or deny this, by looking at the trace json.. don't even need to see logs to tell if something left unfinished was a problem
shengjiz
@shengjiz
Hello, I have a problem, there is one trace between two nodes : SpanA start in nodeA , SpanB start in nodeB after SpanA start , but the startTime of SpanB is smaller than startTime of SpanA in zipkin UI .
@adriancole I think the reason is that the clock between nodes is different, how can i resolve it ?
Adrian Cole
@adriancole
you can use something like NTP to keep your servers clocks in sync
Anders Clausen
@AndersClausen
Just out of curiosity - does Zipkin work with Active Directory? I'm thinking about how to access the dashboard in the cloud and one secure way would be for the users to log in with AD. Any ideas?
Chris K Wensel
@cwensel
sorry if this has been addressed before. but i’m seeing two SERVER spans for a given CLIENT span, making 3 spans, all with the same span ids. but the SERVER span meta-data is different. My guess is one of the SERVER spans is really a child span of the real SERVER span, but something went haywire. known issue?
sun12002
@sun12002
RabbitMQ collector for Zipkin doesn't have any option/configuration for setting a prefetch count on channel. What are the steps/guidelines for proposing a change?
Chris K Wensel
@cwensel
looking at one of the examples of this I found in our data, it feels like we are nesting CLIENT/SERVER spans like this: CLIENT0->CLIENT1->SERVER1->SERVER0. I say this because CLIENT0 is showing up as the parent to the other three spans, and it has no timestamp value.
Adrian Cole
@adriancole
@AndersClausen definitely sounds like a job for a proxy
@cwensel is the parentage CLIENT0->CLIENT1->SERVER1->SERVER0
or is it like CLIENT0->SERVER0, CLIENT0->CLIENT1->SERVER1
the latter is explainable when the async callback has the wrong span on it
which happens to be something tidying today
Chris K Wensel
@cwensel
@adriancole actually CLIENT2,SERVER1,SERVER0 all point to CLIENT0 as the parent, and CLIENT0 has no timestamp. if I sort the available timestamps, they order like CLIENT1->SERVER1->SERVER0, so I assumed CLIENT0 (with no timestamp) came first as its the parent of all of them.
Adrian Cole
@adriancole
well first no timestamp is a bug
no duration can be accidental, but no timestamp is a bug
server1 being a parent of server0 that sounds odd..
can you clean PII from the trace? and/or send to me
Chris K Wensel
@cwensel
client0 is the parent of everyone
(except itself)
Adrian Cole
@adriancole
root span?
or local root in that process
Chris K Wensel
@cwensel
there is, this is just a sub-graph of the trace
not sure about local root.. i’m just looking at raw data, i dont have the code available to me
(raw data before it gets to zipkin server)
Adrian Cole
@adriancole
I mean client0 shows parentID?
Chris K Wensel
@cwensel
to summarize, i have a sub-graph of 4 spans.. client0,client1,server0,server1. client0 is the parent to the others.. client0 has no timestamp. and client1,server1,server0 have timestamps in that order
Adrian Cole
@adriancole
can you turn client0,client1,server0,server1 into (span,localEndpoint) tuples for me?
trying to understand the remote relationships if any
so far I think it looks like a retry
though I can't explain client1, we could ignore it and then it looks like a retry
instrumentation should create a new kind=client (with a new span ID) for each try
but depending on how written, or limits of the framework, you can end up in this situation. in the UI and also the dep linker, we consider the localEndpoint to differentiate when the span IDs are the same
shengjiz
@shengjiz
I read the source code and found that microseconds are calculated by the local relative time(A trace on each node has a relative time), but the relative time of nodes is different, so whether the microseconds calculated at different nodes are comparable.
Adrian Cole
@adriancole
@shengjiz no they aren't strictly comparable. hierarchy can help determine which happened before the other in absolute time. in any case, keeping server times close is helpful
DDd
@webmutation
Hi, is there any up to date Zipkin Helm chart? I could only find the one from Financial Times, but that one seems outdated
I saw Jaeger helm in incubator and was surprised to not see one for zipkin
Adrian Cole
@adriancole
@webmutation dare you to make a pull request to the incubator with a zipkin one :D
DDd
@webmutation
if there is not one more up to date than the financial times one, I will look at their and see if there are many changes needed or not.