Ruby net / ssh与特定主机挂起

Question

I have an issue where if I'm trying to create a ssh connection to a specific host, the connection hangs without timing out. I have tried connecting to this host from the same machine from outside the rails console and it works so I'm assuming this shouldn't be anything related to routing/firewall. I can also confirm I have several other hosts with the exact same OS configuration in different places and they work.

This is the log when running Net::SSH.start in debug mode:

ssh = Net::SSH.start("1.2.3.4", "user", password: "password", verbose: :debug)
2014-02-27 13:17:43 +0100 [DEBUG] establishing connection to 1.2.3.4:22
2014-02-27 13:17:43 +0100 [DEBUG] connection established
2014-02-27 13:17:43 +0100 [INFO] negotiating protocol version
2014-02-27 13:17:43 +0100 [DEBUG] remote is `SSH-2.0-OpenSSH_4.3'
2014-02-27 13:17:43 +0100 [DEBUG] local is `SSH-2.0-Ruby/Net::SSH_2.6.8 i686-linux'
2014-02-27 13:17:43 +0100 [DEBUG] read 704 bytes
2014-02-27 13:17:43 +0100 [DEBUG] received packet nr 0 type 20 len 700
2014-02-27 13:17:43 +0100 [INFO] got KEXINIT from server
2014-02-27 13:17:43 +0100 [INFO] sending KEXINIT
2014-02-27 13:17:43 +0100 [DEBUG] queueing packet nr 0 type 20 len 1620
2014-02-27 13:17:43 +0100 [DEBUG] sent 1624 bytes
2014-02-27 13:17:43 +0100 [INFO] negotiating algorithms
2014-02-27 13:17:43 +0100 [DEBUG] negotiated:
* kex: diffie-hellman-group-exchange-sha1
* host_key: ssh-rsa
* encryption_server: aes128-cbc
* encryption_client: aes128-cbc
* hmac_client: hmac-sha1
* hmac_server: hmac-sha1
* compression_client: none
* compression_server: none
* language_client: 
* language_server: 
2014-02-27 13:17:43 +0100 [DEBUG] exchanging keys
2014-02-27 13:17:43 +0100 [DEBUG] queueing packet nr 1 type 34 len 20
2014-02-27 13:17:43 +0100 [DEBUG] sent 24 bytes

At this point the ssh connection just hangs and could stay like this for 15-30 minutes. Unfortunately I have no error message or anything so I'm really clueless about what the problem might be.

Some specs:

ruby-2.0.0-p0
rails (3.2.13)
net-ssh-2.8.0

The IP address in the log is not a real IP on purpose.

Any suggestion about what the problem could be? Or maybe some other log or place I could check out? I found a similar problem outside SO but it didn't get a solution so I'm trying to ask here...

Answer 1

I solved by reducing maximum transmission unit (MTU). My environment was some specific case. I was trying ssh from VMware Virtual Machine to Openstack Instance. And openstack needed smaller packet to be connected by kitchen-openstack which is using fog , using NET::SSH. Not sure if this works for you, but have a try for these commands (assuming in Ubuntu):

1. check your MTU

   sudo netstat -i

2. You'll get some output like this, in the second column, you can check MTU:

   Iface MTU Met RX-OK RX-ERR RX-DRP RX-OVR TX-OK TX-ERR TX-DRP TX-OVR Flg

   eth0 1500 0 9658 0 0 0 308 0 0 0 BMRU

   lo 16436 0 12952 0 0 0 12952 0 0 0 LRU

3. For network interface eth0, you can try to reduce MTU from 1500 to, for example 1400 like this:

   sudo ifconfig eth0 mtu 1400

4. try net ssh

   ssh = Net::SSH.start("1.2.3.4", "user", password: "password", verbose: :debug)