[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Qemu-discuss] User-mode networking messes up instance server connectabi

From: Alexey Blinov
Subject: [Qemu-discuss] User-mode networking messes up instance server connectability
Date: Mon, 29 Jun 2015 05:21:49 +0300


I'm a newbie to QEMU. I cannot connect to a (simplistic, netcat-based)
TCP ping server on an instance. I guess I'm missing something (problem
seems basic to me). Could anyone help please?

Two instances of RIP linux iso (Recovery Is Possible, 13.7), one NIC at each.

With the user-mode net:
$ qemu-system-x86_64 -machine accel=kvm -m 512 -net
nic,macaddr=52:54:00:00:00:01 -net vde,sock=/tmp/myswitch -net user
-cdrom /path/RIPLinuX-13.7.iso -name t1  &
Second instance is same except mac addr; $ vde_switch -F -sock /tmp/myswitch.
Auto-configured network on both: # dhcpup -u

TCP ping server on the .15: # yes a | nc -l -i 1 -v -n -p 1234
client on .16: # nc -v -n 1234

The client writes error, either "(UNKNOWN) [] 1234 (?) open"
or "(UNKNOWN) [] 1234 (?) : Connection reset by peer" (each
with about 0.5 probablility), and returns. The server doesn't show
anything. Normal ICMP pings of course fly fine.

Without user-mode net (removed "-net user"): manually configured
"ifconfig eth0 up" (on 1st, and similarly except .16 on
2nd). Works fine (client writes line "a" each second).

qemu-x86_64 version 2.1.2, on Debian Jessie (systemd removed).

I have this problem only when connecting to a server on an instance.
Connects to real external servers (ssh,http) are fine. Similar UDP
netcat server (the "-u" option) used to work (with some qemu
options/disk images), but as I just tried - doesn't either.

(Of course, I have another server to which I cannot connect, and,
trying to figure it out, I simplified it to netcat).


reply via email to

[Prev in Thread] Current Thread [Next in Thread]