tripleslash
tripleslash
I had a similar issue once in an implementation of a TCP SSL socket using asio. Turned out I had to call asio::ssl_stream::shutdown before calling the socket close method. Maybe...
Also happens with latest trantor (v1.5.15) and drogon (v1.9.1)
Custom built Linux BSP with Kernel 6.5.12 and OpenSSL 3.1.0. I cant change my TLS Provider because many other applications on my system depend on OpenSSL Compiler is GCC 12.2.1...
``` 20231128 08:24:30.062488 UTC 4347 TRACE [newFileResponse] send http file:/var/www//ui/assets/images/loadingd80ae7c176ae41d8797c.png offset 0 length 0 - HttpResponseImpl.cc:296 20231128 08:24:30.063095 UTC 4347 TRACE [sendStaticFileResponse] Save in cache for 5 seconds - StaticFileRouter.cc:562...
`TRACE [sendFileInLoop] send file in loop: break on socket buffer full (?) - TcpConnectionImpl.cc:1215` this line always seems to happen when the issue appears
`Resource temporarily unavailable (errno=11) send file in loop: break on socket buffer full (?) - 11` errno = 11 = EAGAIN
> That is normal and expected. That sounds like the async socket is full and we are waiting for the OS to make the socket writable again. Hmm.... I agree...
The issue seems to happen more frequently when you try to connect to the device from outside the intranet or over SSH proxies for example. In intranet it is hard...
> @tripleslash I've looked into the issue but it's baffling me. Is it possible for me to get access to a system that can replicate the issue? > > Another...
@marty1885 This issue still persists with the current master of trantor+drogon