-
Notifications
You must be signed in to change notification settings - Fork 2.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Bootstrap discovery (locating config server via discovery) leads to harmless RejectedExecutionException in logs #1055
Comments
I built a setup with 2 peer aware eureka servers and one eureka client with the configuration you provided. I am not reproducing your problem. Can you upload to GH a code snippet reproducing this problem ? |
I am also concerned by your stacktrace.
It feels to me that you have overwritten eureka thread pool configuration. With this setup, any task will fail since your pool size are defined to 0 according to your stack trace. |
@daniellavoie I'll prepare a sample project along the day. |
I have finally prepared the sample projects: https://github.com/codependent/scn-brixton-release.git Actually it has nothing to do with using peer Eureka servers, the miminum scenario needed is one Eureka Server, one config server and one microservice (Eureka client that uses the config server). When I remove the config server the error doesn't show up. To reproduce just start the Eureka server (archimedes), when it's up, start the Config server, and when everything is up just launch the user-api server. You'll see the usual:
Then after 5 minutes the error shows up in the log:
|
Maybe i'm wrong, but i've seen this error when you have the devtools on the classpath.
when I recompile in intellij a lot of my apps show this error and I have to restart them manually. Idk if it's can be you're case! |
@royremi I just checked the dependencies. I don't have it, not even as a transitive dependency. |
Great, got to reproduce it. I'll debug to see what is wrong with that executor. |
I won't have too much time for the rest of the week but here is what I found for now. There is two discovery client running. Spring Cloud AutoConfiguration sends a termination signals to the EurekaClient located in the parent spring context. This signal terminates the executor service who receives a new task from TimedSupervisorTask. Since the EurekaClient was previously shutdown, the tasks fails and generate the stack trace. I haven't figured out why the TimedSupervisorTask is generated for that terminated client. I'm not sure if it is normal to have 2 discovery clients running. One comes from the bootstrap mechanism. |
I was on Brixton.M3 and there were exceptions reported as in Netflix/eureka#638. Now I upgraded but have a different exception reported in this issue. Given the simple reproduction of the latest exception, I'm wondering how did any test app manage to not throw this exception during the GA release of Brixton. What does the Spring Cloud team do differently (for integration testing) that we may be missing? |
I accidentally had this in my Microservice's .yml file:
After reading this bug report, I wondered if perhaps "eureka->instance" variables shouldn't exist in a Non-Eureka project. So I moved that configuration to my Eureka project, and it seems to be happy. EDIT: I take that back. The error still occurs for me. |
same error when I change my service to use spring cloud config discovery in bootstrap.yml
in my service configuration, I also use EnableDiscoveryClient annotation looks like some of the logs are duplicate
|
My "solution" for this error: logging:
level:
com.netflix.discovery.TimedSupervisorTask: "OFF" |
You call that a solution? That's like doctor telling a patient not to worry because the MRI machine isn't working. |
@abhijitsarkar please be aware of your tone. That last comment was not necessary. So far, besides the error showing in the log, no one has demonstrated that it is negatively affecting their applications. |
My tone was humorous with a hint of sarcasm. Besides, this is an open forum where people are discussing issues about a software. If you expect everyone to always speak in a manner that's pleasing to you, consider moving to a closed forum like Spring JIRA with only developer access. |
Tone and sarcasm don't always come across really well in short text. I wasn't the only one who reacted like I did. We have a Code of Conduct we expect folks to abide by. |
Pleasing to me vs. respectful to all is very different. |
My analogy to the doctor in response to a suggestion of suppressing error logs was disrespectful? And violates the code of conduct? Sorry, I just don't get it. The last relevant comment from a committer was 3 months ago saying that they won't have time to look into this issue. And from your comment above, it seems like you're waiting for the users to prove that they are negatively impacted before you do anything about it. Of course, this is a community driven open source but don't you think ludicrous log suppression isn't the answer? |
How to reproduce the error: With this Eureka configuration:
This POM configuration;
This configuration client:
The error:
Versions:
Upgrade the Netflix version, shows the log in warn level, is that the solution for this problem? Thanks. |
Thanks. Please create a sample project and link to it. And please format your code snippets. How did you upgrade the netflix version? What was the outcome? |
FWIW I updated the app in the @codependent sample above to use Dalston.SR1 and I don't see any exceptions logged at all (after running for about 10min). 5min seems to be the pattern with the original sample (Brixton.RELEASE). I propose we close this as fixed. |
Hi, thanks for your reply. I've tested the @codependent and with the original dependencies (Brixton) the error shows after 5 minutes. With Camdem.SR4 the error shows again. And with Dalston.SR1 the error shows again.
I've updated this dependencies:
|
Dalston is not compatible with boot 1.4.x, btw. |
You're right, but with the 1.5.x the error continues :( |
Not for me. What log level is it coming out at? What did you do to see that log? You didn't even paste the whole stack trace BTW. |
I just updated the sample proyect to work with Spring Boot 1.5.4.RELEASE and Spring Cloud Dalston.SR1. With this setup the exception no longer shows up. Thank you for having a look at it. |
Hi @codependent , with Spring Boot 1.5.4.RELEASE the exception dissapear, thanks. |
I observed if i set eureka.client.enabled=true in my config server and
Versions:
|
@PANDEYmradul this issue is closed and yours is unrelated. Please open another ticket. |
My microservices' logs show periodically, can't tell when, just at a certain moment the following exception:
My infraestructure has two Eureka peer aware servers, with the following configuration:
All my services, at some moment throw that exception. This is and example of one of them:
Despite the exception everything keeps running OK, but it's kind of annoying to get it. If you think this should be a Netflix's Eureka issue let me know and I'll close this one.
The text was updated successfully, but these errors were encountered: