APM Agent language and version:
dotnet 1.8.0+8605545e8cbc5b7d68941354a22099a9c43a4aca
Browser version:
Same problem with Chrome 98.0.4758.82 as with MS Edge 98.0.1108.43 and also same with Firefox 97.0
Original install method (e.g. download page, yum, deb, from source, etc.) and version:
The APM agent is deployed as part of our CI/CD build+deploy process. Code change has to be done in order to send the metrics to elastic apm server. App team included Elastic.Apm.NetCoreAll" Version="1.8.0" and "Elastic.Apm.SerilogEnricher" Version="1.5.1"
Fresh install or upgraded from other version?
Fresh installation
Is there anything special in your setup?
Direct communication from APM agent -> APM server.
**Description of the problem including expected versus actual behavior. **:
This is a fairly new environment for us, we need to learn how to tune the instrumentation in a way which populates data in the Dependencies view. We have got no data in them; neither from Overview->Dependencies nor from FocusMap->Dependencies.
We do have good data in Service Map, Services, Traces, Transactions.
But not in Dependencies.
Currently I suspect we are missing instrumentation for the integration request, which is using graphQL requests - we use some "GraphQL.Client" nuget-package, to send requests to our Federated Graph integration platform.
New info: The GraphQL client ... graphql-dotnet/graphql-client: A GraphQL Client for .NET Standard (github.com) .... is perhaps not instrumented by our current .... Elastic.Apm.NetCoreAll" Version="1.8.0 ?
Provide logs and/or server output (if relevant):
Which logs should I look for here, in this case of troubleshooting? Any hints are appreciated.
from the .NET Agent's perspective I can tell that there is no specific instrumentation for graphql-dotnet/graphql-client, but the agent would still recognise those calls as an outgoing HTTP call, so you'd see those in the Dependency view.
Overall to me this seems more like either an issue in Kibana or in Elasticsearch. I say this based on the fact that you say that "We do have good data in Service Map, Services, Traces, Transactions.". If you see the outgoing HTTP calls to the GraphQL service, then the same data would be used to populate the dependencies view.
Which logs should I look for here, in this case of troubleshooting? Any hints are appreciated.
I think Kibana logs would help here as the first step - and then I'd ask people from the APM UI team to look at it.
I'd also note that the APM .NET Agent version 1.8.0 is a bit old now, we had multiple releases since then, but I don't think it'd change much on this specific issue if you'd update to the latest Agent version.
Hi, thanks!
Elastic Support gave advice for where to look to verify data ... and we miss data in the related APM metric indicies. Yet we do not know why, but Support will discuss with APM dev team.
Would you know a way to send command line commands to an Agent running in Docker, in k8 cluster (running together with the app it monitors) ?
Do you mean setting agent configs? Like from this list?
If so, then yes, that's possible. You have 2 options, it depends on how you configure the agent:
If you pass the IConfiguration instance to the agent (e.g. when you enable the agent like app.UseAllElasticApm(Configuration)) then you just need to add the command line config source to the IConfiguration - docs here). In that setup the agent reads config from the IConfiguration instance, so whatever config source you add to it, the agent will just use it and read form it.
If you enabled the agent differently (so no IConfiguration was passed to it`) then it reads configs from environment variables. It's not really reading from command line, but I guess you can achieve the same result. So in that case, just set the corresponding environment variable and the agent will pick it up.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.