Something makes curl hang for 5 minutes after connection refused

[Basin Ilya]

Hi.

Recently I noticed that `curl` does not fail immediately after connection refused, but waits for 5 minutes.

This only happens on my machine and only with the `Cygwin` version of `curl`. The mingw version is unaffected.
I tried downgrading curl, but it didn't help. I thought there's a rogue setting in `/etc` or in home or maybe some environment variable, but I couldn't find it.

    ***@BASIN /cygdrive/c/Users/basin
    $ >/dev/null /usr/bin/curl -v http://127.0.0.1:2/
    * STATE: INIT => CONNECT handle 0x80048658; line 1404 (connection #-5000)
    * Added connection 0. The cache now contains 1 members
    *   Trying 127.0.0.1...
    * TCP_NODELAY set
    * STATE: CONNECT => WAITCONNECT handle 0x80048658; line 1456 (connection #0)
      % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                     Dload  Upload   Total   Spent    Left  Speed
      0     0    0     0    0     0      0      0 --:--:--  0:04:59 --:--:--     0* Connection timed out after 300145 milliseconds
    * multi_done
    * stopped the pause stream!
      0     0    0     0    0     0      0      0 --:--:--  0:05:00 --:--:--     0
    * Closing connection 0
    * The cache now contains 0 members
    curl: (28) Connection timed out after 300145 milliseconds
   
    ***@BASIN /cygdrive/c/Users/basin
    $ /usr/bin/curl --version
    curl 7.59.0 (i686-pc-cygwin) libcurl/7.59.0 OpenSSL/1.0.2p zlib/1.2.11 libidn2/2.0.4 libpsl/0.18.0 (+libidn2/2.0.2) libssh2/1.7.0 nghttp2/1.31.0
    Release-Date: 2018-03-14
    Protocols: dict file ftp ftps gopher http https imap imaps ldap ldaps pop3 pop3s rtsp scp sftp smb smbs smtp smtps telnet tftp
    Features: AsynchDNS Debug IDN IPv6 Largefile GSS-API Kerberos SPNEGO NTLM NTLM_WB SSL libz TLS-SRP HTTP2 UnixSockets HTTPS-proxy PSL Metalink

Can someone try to reproduce it?


--
Problem reports: http://cygwin.com/problems.html
FAQ: http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple

[Lee]

Post by Basin Ilya
Hi.
Recently I noticed that `curl` does not fail immediately after connection
refused, but waits for 5 minutes.
This only happens on my machine and only with the `Cygwin` version of
`curl`. The mingw version is unaffected.
I tried downgrading curl, but it didn't help. I thought there's a rogue
setting in `/etc` or in home or maybe some environment variable, but I
couldn't find it.
$ >/dev/null /usr/bin/curl -v http://127.0.0.1:2/
* STATE: INIT => CONNECT handle 0x80048658; line 1404 (connection
#-5000)
* Added connection 0. The cache now contains 1 members
* Trying 127.0.0.1...
* TCP_NODELAY set
* STATE: CONNECT => WAITCONNECT handle 0x80048658; line 1456 (connection
#0)

<.. snip ..>

Post by Basin Ilya
curl: (28) Connection timed out after 300145 milliseconds
$ /usr/bin/curl --version
curl 7.59.0 (i686-pc-cygwin) libcurl/7.59.0 OpenSSL/1.0.2p zlib/1.2.11
libidn2/2.0.4 libpsl/0.18.0 (+libidn2/2.0.2) libssh2/1.7.0 nghttp2/1.31.0
Can someone try to reproduce it?

I get the same behavior:
$ /usr/bin/curl -v http://127.0.0.1:2/
* STATE: INIT => CONNECT handle 0x600057ad0; line 1404 (connection #-5000)
* Added connection 0. The cache now contains 1 members
* Trying 127.0.0.1...
* TCP_NODELAY set
* STATE: CONNECT => WAITCONNECT handle 0x600057ad0; line 1456 (connection #0)
* Connection timed out after 300324 milliseconds
* multi_done
* stopped the pause stream!
* Closing connection 0
* The cache now contains 0 members
curl: (28) Connection timed out after 300324 milliseconds

$ /usr/bin/curl --version
curl 7.59.0 (x86_64-unknown-cygwin) libcurl/7.59.0 OpenSSL/1.0.2p
zlib/1.2.11 libidn2/2.0.4 libpsl/0.18.0 (+libidn2/2.0.2) libssh2/1.7.0
nghttp2/1.31.0


I also get an almost immediate 'failed to connect' notice on windows:
C:\>C:\UTIL\curl\curl.exe -v http://127.0.0.1:2/
* Trying 127.0.0.1...
* TCP_NODELAY set
* connect to 127.0.0.1 port 2 failed: Connection refused
* Failed to connect to 127.0.0.1 port 2: Connection refused
* Closing connection 0
curl: (7) Failed to connect to 127.0.0.1 port 2: Connection refused

with either version of curl:
C:\>where curl
C:\UTIL\curl\curl.exe
C:\Windows\System32\curl.exe

Regards,
Lee

--
Problem reports: http://cygwin.com/problems.html
FAQ: http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple

[Brian Inglis]

Post by Lee

Post by Basin Ilya
Hi.
Recently I noticed that `curl` does not fail immediately after connection
refused, but waits for 5 minutes.
This only happens on my machine and only with the `Cygwin` version of
`curl`. The mingw version is unaffected.
I tried downgrading curl, but it didn't help. I thought there's a rogue
setting in `/etc` or in home or maybe some environment variable, but I
couldn't find it.
$ >/dev/null /usr/bin/curl -v http://127.0.0.1:2/
* STATE: INIT => CONNECT handle 0x80048658; line 1404 (connection
#-5000)
* Added connection 0. The cache now contains 1 members
* Trying 127.0.0.1...
* TCP_NODELAY set
* STATE: CONNECT => WAITCONNECT handle 0x80048658; line 1456 (connection
#0)

<.. snip ..>

Post by Basin Ilya
curl: (28) Connection timed out after 300145 milliseconds
$ /usr/bin/curl --version
curl 7.59.0 (i686-pc-cygwin) libcurl/7.59.0 OpenSSL/1.0.2p zlib/1.2.11
libidn2/2.0.4 libpsl/0.18.0 (+libidn2/2.0.2) libssh2/1.7.0 nghttp2/1.31.0
Can someone try to reproduce it?

$ /usr/bin/curl -v http://127.0.0.1:2/
* STATE: INIT => CONNECT handle 0x600057ad0; line 1404 (connection #-5000)
* Added connection 0. The cache now contains 1 members
* Trying 127.0.0.1...
* TCP_NODELAY set
* STATE: CONNECT => WAITCONNECT handle 0x600057ad0; line 1456 (connection #0)
* Connection timed out after 300324 milliseconds
* multi_done
* stopped the pause stream!
* Closing connection 0
* The cache now contains 0 members
curl: (28) Connection timed out after 300324 milliseconds
$ /usr/bin/curl --version
curl 7.59.0 (x86_64-unknown-cygwin) libcurl/7.59.0 OpenSSL/1.0.2p
zlib/1.2.11 libidn2/2.0.4 libpsl/0.18.0 (+libidn2/2.0.2) libssh2/1.7.0
nghttp2/1.31.0
C:\>C:\UTIL\curl\curl.exe -v http://127.0.0.1:2/
* Trying 127.0.0.1...
* TCP_NODELAY set
* connect to 127.0.0.1 port 2 failed: Connection refused
* Failed to connect to 127.0.0.1 port 2: Connection refused
* Closing connection 0
curl: (7) Failed to connect to 127.0.0.1 port 2: Connection refused
C:\>where curl
C:\UTIL\curl\curl.exe
C:\Windows\System32\curl.exe

Ditto all the way!

Could this be caused by network speed improvements made to Cygwin a while back?
Or were those patches never sent/received/applied from Daniel Havey (MS Windows
Program Manager for Transports and IP)?
--
Take care. Thanks, Brian Inglis, Calgary, Alberta, Canada

This email may be disturbing to some readers as it contains
too much technical detail. Reader discretion is advised.

--
Problem reports: http://cygwin.com/problems.html
FAQ: http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple

[Corinna Vinschen]

Post by Brian Inglis

Post by Lee

Post by Basin Ilya
Hi.
Recently I noticed that `curl` does not fail immediately after connection
refused, but waits for 5 minutes.
This only happens on my machine and only with the `Cygwin` version of
`curl`. The mingw version is unaffected.
I tried downgrading curl, but it didn't help. I thought there's a rogue
setting in `/etc` or in home or maybe some environment variable, but I
couldn't find it.
$ >/dev/null /usr/bin/curl -v http://127.0.0.1:2/
* STATE: INIT => CONNECT handle 0x80048658; line 1404 (connection
#-5000)
* Added connection 0. The cache now contains 1 members
* Trying 127.0.0.1...
* TCP_NODELAY set
* STATE: CONNECT => WAITCONNECT handle 0x80048658; line 1456 (connection
#0)

<.. snip ..>

Post by Basin Ilya
curl: (28) Connection timed out after 300145 milliseconds
$ /usr/bin/curl --version
curl 7.59.0 (i686-pc-cygwin) libcurl/7.59.0 OpenSSL/1.0.2p zlib/1.2.11
libidn2/2.0.4 libpsl/0.18.0 (+libidn2/2.0.2) libssh2/1.7.0 nghttp2/1.31.0
Can someone try to reproduce it?

$ /usr/bin/curl -v http://127.0.0.1:2/
* STATE: INIT => CONNECT handle 0x600057ad0; line 1404 (connection #-5000)
* Added connection 0. The cache now contains 1 members
* Trying 127.0.0.1...
* TCP_NODELAY set
* STATE: CONNECT => WAITCONNECT handle 0x600057ad0; line 1456 (connection #0)
* Connection timed out after 300324 milliseconds
* multi_done
* stopped the pause stream!
* Closing connection 0
* The cache now contains 0 members
curl: (28) Connection timed out after 300324 milliseconds
$ /usr/bin/curl --version
curl 7.59.0 (x86_64-unknown-cygwin) libcurl/7.59.0 OpenSSL/1.0.2p
zlib/1.2.11 libidn2/2.0.4 libpsl/0.18.0 (+libidn2/2.0.2) libssh2/1.7.0
nghttp2/1.31.0
C:\>C:\UTIL\curl\curl.exe -v http://127.0.0.1:2/
* Trying 127.0.0.1...
* TCP_NODELAY set
* connect to 127.0.0.1 port 2 failed: Connection refused
* Failed to connect to 127.0.0.1 port 2: Connection refused
* Closing connection 0
curl: (7) Failed to connect to 127.0.0.1 port 2: Connection refused
C:\>where curl
C:\UTIL\curl\curl.exe
C:\Windows\System32\curl.exe

Ditto all the way!
Could this be caused by network speed improvements made to Cygwin a while back?
Or were those patches never sent/received/applied from Daniel Havey (MS Windows
Program Manager for Transports and IP)?

Somebody may want to bisect this...


Corinna