epoll | Event-based socket server , which uses epoll | Socket library

When I run bundle install on my rails 5 project, I am getting a number of errors on gems that appear to have native extensions. Here is the top of the output for one of those gems:

Is it possible to use http 1.1 keep-alive with the boost beast library when using a sync ssl client? I have a process that works something like this:

1. Connection to web server using https. Send an initial request, wait for a response
2. select()/epoll()/etc on the https native socket handle
3. When the native handle has something ready, I issue an http::read() on the ssl_stream.

The first http::read() works perfectly fine. However, when the socket is ready the next time, I get an end of stream exception when issuing an http::read() on the socket. What I want is for the socket to stay alive for multiple requests.

The code isn't too complicated, but it's in pieces:

I am running a Spring Boot app that uses WebClient for both non-blocking and blocking HTTP requests. After the app has run for some time, all outgoing HTTP requests seem to get stuck.

WebClient is used to send requests to multiple hosts, but as an example, here is how it is initialized and used to send requests to Telegram:

WebClientConfig:

Env:

After upgrading from spring-boot-parent version 2.5.5 to 2.6.0, I started seeing these error messages spamming the logs:

[INFO] [stdout] 2022-01-11 13:40:01.157 WARN 76859 --- [ udp-epoll-2] i.m.s.reactor.netty.channel.FluxReceive : [6d1243de, L:/127.0.0.1:58160 - R:localhost/127.0.0.1:8125] An exception has been observed post termination, use DEBUG level to see the full stack: java.net.PortUnreachableException: readAddress(..) failed: Connection refused

Using DEBUG level:

I receive sporadic the follow error with the Azure CosmosDB emulator on calling CosmosDatabase.createContainerIfNotExists.

• Also if the error message contains many helpless details there is nothing to understand the cause. How can I get more information about the error from the Azure CosmosDB emulator?

• What can be the cause of the error? The count of existing container if the error occur is approximate 70 to 130. This should be no problem for the limit of 250 containers.

• What is the best handling of the error?

I have a program which reads from a file which is a list of domain names. It performs asynchronous DNS and then downloads the landing page for each domain using an asynchronous epoll loop.

The program runs fine for thousands of iterations and then bombs out with a *** buffer overflow detected ***: terminated error. Here is the backtrace:

Provided that:

• I have created an epoll instance epfd by epoll_create, and registered many regular file descriptors by EPOLL_CTL_ADD.
• I want to close the epoll instance by close(epfd)

The manual page does't say whether I must EPOLL_CTL_DEL all file descriptors before close(epfd).

So, my question is:

Is it necessary to remove all file descriptors in the interest list before closing the epoll instance itself?

This is my code: