-
Luca, thank you for the quick answer. Till the fix is available, I'll take your suggestions and change my scripts.
2009-11-21 17:03:30 UTC by frankpelosato
-
How to reproduce:
rm -rf /tmp/switch
vde_switch -s /tmp/switch/data
In this case vde_switch will create the control socket in /tmp/switch/ctl instead of /tmp/switch/data/ctl .
The expected behavior is either to fail startup or to recursively create non-existing directories.
The bug seems to be around vde-2/src/vde_switch/datasock.c:367 during vde_realpath() call.
2009-11-21 00:48:52 UTC by shammash
-
It seems there's a bug in the path canonicalization. I'm opening a bug for that and I'll try to fix it ASAP.
As a workaround just add "mkdir -p /var/run/vde2/switch1" before starting the switch.
On a side note: instead of killing the switch you can also use "vdecmd -s /var/run/vde2/switch0/mgmt shutdown".
Thank you,
Luca.
2009-11-21 00:35:05 UTC by shammash
-
Hello together,
yesterday, I started using vde on my Ubuntu KARMIC box. I start 2 vde switches via upstart (the details are listed in the protocol).
The following command will be used within the scripts:
/usr/bin/vde_switch --sock /var/run/vde2/switch0/ctl \
--mode 666 \
--dirmode 777 \
--pidfile /var/run/vde2/switch0/pid \
--mgmt /var/run/vde2/switch0/mgmt \...
2009-11-20 15:50:12 UTC by frankpelosato
-
fgiunchedi committed revision 401 to the VDE: Virtual Distributed Ethernet SVN repository, changing 11 files.
2009-10-17 21:29:00 UTC by fgiunchedi
-
Hi,
I believe the bug has been fixed in svn revision 400.
Thank you,
Luca.
2009-10-01 20:56:49 UTC by shammash
-
shammash committed revision 400 to the VDE: Virtual Distributed Ethernet SVN repository, changing 1 files.
2009-10-01 20:53:54 UTC by shammash
-
Yes,
the received packet has 4 bytes more and it's also forward-shifted by 4 bytes
(so you get nonsense MAC addresses).
Will try to understand why in the next days, I hope..
Luca.
2009-10-01 17:34:10 UTC by shammash
-
Hi.
See below the reproduction steps. configuration on the second computer is similar (IP changed, and no dpipe).
Looking at counters on second switch, I could see the counters increase. So I did a tcpdump on the second computer's tap1 interface :
When pdump is activated, received frames are 4 bytes longer than expected (I got 4 bytes more after ping data), which explains why the request gets...
2009-09-30 08:38:12 UTC by nobody
-
Hi,
please can you check if ports packets/bytes counters are updated and if pdump dumps something in its file/pipe?
Can you also write here the commands/config-files you use?
Thank you,
Luca.
2009-09-29 18:42:13 UTC by shammash