sshd (and other daemons) on OpenVZ HN resets connection [message #34641] |
Thu, 22 January 2009 19:51 |
Sixtus
Messages: 7 Registered: January 2009 Location: Germany
|
Junior Member |
|
|
Hi all,
my first post and question.
First my system:
OpenVZ running on DELL 2950 under Etch, kernel 2.6.18-14-fza-amd64, installed from repository.
4 network interfaces, OpenVZ is running on eth0 which is currently the only connected interface.
IP address is 10.150.127.61/255.255.224.0 so a pure internal network.
VEs have have IPs of the range 10.150.103.0/255.255.224.0 and work perfectly.
Well, sshd (and some other daemons) run on eth0 and they do not work well.
After rebooting the HN, everything is fine but if I start to copy a lot of files from another server in the network to a VE, and if I am connected to the HN via ssh, sshd tells me "connection reset by peer". Also another daemon, running on eth0 (snmpd) is unavailable.
Wierd thing: sshd on the HN is still accessible from a VE via 10.150.127.61.
But if no VE runs I have no chance to log in to the HN (and have to walk thru the building to the server room).
So I am sure there is a misconfiguration in the system but I have no idea where to look for.
Btw: IP forwarding is activated.
I hope someone will give me a hint
Thanks and best regards,
Sixtus
|
|
|
|
|
Re: sshd (and other daemons) on OpenVZ HN resets connection [message #34661 is a reply to message #34641] |
Sat, 24 January 2009 08:57 |
n00b_admin
Messages: 77 Registered: July 2006 Location: Romania
|
Member |
|
|
Ok, please correct me if i misunderstood your setup.
You are having problems with a ssh session from an external machine to a VE inside the HN or the HN itself ?
If you are doing a ssh to a VE check your beancounters for fail counts.
I had problems related to connection resets for VEs that were configured with more strict beancounter values by default.
Web traffic would fail with connection resets due to exceeded limits for tcp socket buffers.
|
|
|