Xen users - May 2010 - /etc/xen/scripts/network-bridge: line 118: sigerr: command not found

Hi there,

after a xen upgrade (to version 4.0.0) I get the message in the Subject when
xend daemon tries to start.
My system is Gentoo Linux 64bit kernel xen-sources-2.6.31-r7.
The problem is that the xenbr0 bridge doesn''t start so that the error
occurs.
I''ve just checked the network-bridge option in the
/etc/xen/xend-config.spx
file and it is correct, as well as the network configuration which has been
working since so far:
(network-script network-bridge).

What could be happened?

Thanks,

-- 
Flavio


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

give the output of brctl show.

On Wed, May 26, 2010 at 6:01 PM, Flavio <fbcyborg@gmail.com>
wrote:
> Hi there,
>
> after a xen upgrade (to version 4.0.0) I get the message in the Subject
when
> xend daemon tries to start.
> My system is Gentoo Linux 64bit kernel xen-sources-2.6.31-r7.
> The problem is that the xenbr0 bridge doesn''t start so that the
error
> occurs.
> I''ve just checked the network-bridge option in the
/etc/xen/xend-config.spx
> file and it is correct, as well as the network configuration which has been
> working since so far:
> (network-script network-bridge).
>
> What could be happened?
>
> Thanks,
>
> --
> Flavio
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
>


-- 
Tapas

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Hi Tapas,

thank you for the reply.

On 26 May 2010 18:43, Tapas Mishra <mightydreams@gmail.com> wrote:
> give the output of brctl show.
>
bridge name     bridge id               STP enabled     interfaces
eth0            8000.00609720464b       no              peth0


That''s strange, because, I always used xenbr0 as bridge as specified in
the
/etc/conf.d/net file.
-- 
Flavio


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Actually that is the same problem I faced on my Xen.
I read some where about the working of
 Xen bridge but it does not works they way it is exactly said.
Till today this problem I could not understand.
I did not got time to read the networking scripts on Xen.

On Wed, May 26, 2010 at 10:41 PM, Flavio <fbcyborg@gmail.com>
wrote:
> Hi Tapas,
>
> thank you for the reply.
>
> On 26 May 2010 18:43, Tapas Mishra <mightydreams@gmail.com> wrote:
>>
>> give the output of brctl show.
>
> bridge name     bridge id               STP enabled     interfaces
> eth0            8000.00609720464b       no              peth0
>
>
> That''s strange, because, I always used xenbr0 as bridge as
specified in the
> /etc/conf.d/net file.
> --
> Flavio
>


-- 
Tapas

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On 26 May 2010 20:18, Tapas Mishra <mightydreams@gmail.com> wrote:
> Actually that is the same problem I faced on my Xen.
> I read some where about the working of
>  Xen bridge but it does not works they way it is exactly said.
> Till today this problem I could not understand.
> I did not got time to read the networking scripts on Xen.
>
So, as matter of fact didn''t you solve the problem too?
-- 
Flavio


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Thu, May 27, 2010 at 12:14 AM, Flavio <fbcyborg@gmail.com> wrote:
> So, as matter of fact didn''t you solve the problem too?
> --
> Flavio
>
No  but I also want to add that I am not expert at Xen.
I tried doing  various permutations on Xen but any of them did not worked.


-- 
Tapas

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

I had the same problem on my gentoo xen 4.0 2.6.32 machine.

It happens because the locking did not work in the network scripts of
gentoo. I removed the locking from them and then it worked.

When the locking was used, it waited for 100 times to get it, failed and
then in the line where the default gateway for the xenbr0 was installed
the "ip" command was called with "sigerr" instead of the
default gateway
extracted previously by the ip route command on the original network
interface.

For me, removing of locking in /etc/xen/scripts/network-bridge helped.

Daniel


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On 27 May 2010 08:32, Daniel Schmitt <daniel.schmitt@udo.edu> wrote:
> I had the same problem on my gentoo xen 4.0 2.6.32 machine.
>
> It happens because the locking did not work in the network scripts of
> gentoo. I removed the locking from them and then it worked.
>
> When the locking was used, it waited for 100 times to get it, failed and
> then in the line where the default gateway for the xenbr0 was installed
> the "ip" command was called with "sigerr" instead of
the default gateway
> extracted previously by the ip route command on the original network
> interface.
>
> For me, removing of locking in /etc/xen/scripts/network-bridge helped.
>
Thank you Daniel,

as far as I understand there''s something "wrong" in the
/etc/xen/scripts/network-bridge,
but what, exactly? What do you mean when you say "removing the lock"?
What should I do exactly?

I saw there is a line like this:
claim_lock "network-bridge"
and another like this:
release_lock "network-bridge"
in two functions.

Should they be removed?
-- 
Flavio


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Hi I have a question:

Is NIC passthrough related to this file
"/etc/xen/scripts/network-
>
> bridge"

Thanks
Lei


On Thu, May 27, 2010 at 2:52 PM, Flavio <fbcyborg@gmail.com> wrote:
> On 27 May 2010 08:32, Daniel Schmitt <daniel.schmitt@udo.edu> wrote:
>
>> I had the same problem on my gentoo xen 4.0 2.6.32 machine.
>>
>> It happens because the locking did not work in the network scripts of
>> gentoo. I removed the locking from them and then it worked.
>>
>> When the locking was used, it waited for 100 times to get it, failed
and
>> then in the line where the default gateway for the xenbr0 was installed
>> the "ip" command was called with "sigerr" instead
of the default gateway
>> extracted previously by the ip route command on the original network
>> interface.
>>
>> For me, removing of locking in /etc/xen/scripts/network-bridge helped.
>>
> Thank you Daniel,
>
> as far as I understand there''s something "wrong" in the
> /etc/xen/scripts/network-bridge,
> but what, exactly? What do you mean when you say "removing the
lock"?
> What should I do exactly?
>
> I saw there is a line like this:
> claim_lock "network-bridge"
> and another like this:
> release_lock "network-bridge"
> in two functions.
>
> Should they be removed?
> --
> Flavio
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
>


-- 
"We learn from failure, not from success!"


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On 27 May 2010 08:52, Flavio <fbcyborg@gmail.com> wrote:
> I saw there is a line like this:
> claim_lock "network-bridge"
> and another like this:
> release_lock "network-bridge"
> in two functions.
>
> Should they be removed?
>
I''ve just tried to remove that lines.
That is not the solution, even if the message in the subject
doesn''t longer appear, but the network is still inactive.

-- 
Flavio


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

> as far as I understand there''s something "wrong" in the
> /etc/xen/scripts/network-bridge,
> but what, exactly? What do you mean when you say "removing the
lock"?
> What should I do exactly?
>
> I saw there is a line like this:
> claim_lock "network-bridge"
> and another like this:
> release_lock "network-bridge"
> in two functions.
>
> Should they be removed?
yes, remove all lines containing ''*lock
"network-bridge"'' in
/etc/xen/scripts/network-bridge.
I did not exactly remember the syntax in my last post ...

Daniel


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Workaround.
Comment out all xen bridge declarations in /etc/xen/xend-config.sxp.
Create bridge manually.
View :-

http://wiki.libvirt.org/page/Networking

Boris.

--- On Thu, 5/27/10, Flavio <fbcyborg@gmail.com> wrote:

From: Flavio <fbcyborg@gmail.com>
Subject: Re: [Xen-users] /etc/xen/scripts/network-bridge: line 118: sigerr: 
command not found
To: daniel.schmitt@udo.edu
Cc: xen-users@lists.xensource.com
Date: Thursday, May 27, 2010, 3:26 AM



On 27 May 2010 08:52, Flavio <fbcyborg@gmail.com> wrote:


I saw there is a line like this: 
claim_lock "network-bridge"
and another like this:

release_lock "network-bridge"
in two functions.

Should they be removed?
I''ve just tried to remove that lines.
That is not the solution, even if the message in the subject


doesn''t longer appear, but the network is still inactive.

-- 
Flavio


-----Inline Attachment Follows-----

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users


      

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On 27 May 2010 09:27, Daniel Schmitt <daniel.schmitt@udo.edu> wrote:
>
> yes, remove all lines containing ''*lock
"network-bridge"'' in
> /etc/xen/scripts/network-bridge.
> I did not exactly remember the syntax in my last post ...
>
As I said in the previous message, that didn''t solve my problem.
The message no longer appear but the network doesn''t come up
automatically as well as before.

I''m going to make another attempt.

Thank you.
-- 
Flavio


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

> On 27 May 2010 08:52, Flavio <fbcyborg@gmail.com> wrote:
> I''ve just tried to remove that lines.
> That is not the solution, even if the message in the subject
> doesn''t longer appear, but the network is still inactive.
Hmm.
I found the place where something wrong happened by using bash debug mode.
Add " -x" to the 1st line of network-bridge script. There is something
written "#!/bin/bash". Then you will see a lot of debug output next
boot.

For my problem at one place in the script "sigerr" was returned by one
command/function and incorrectly used as gateway ip address.

Daniel


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On 27 May 2010 09:40, Daniel Schmitt <daniel.schmitt@udo.edu> wrote:
> Hmm.
> I found the place where something wrong happened by using bash debug mode.
> Add " -x" to the 1st line of network-bridge script. There is
something
> written "#!/bin/bash". Then you will see a lot of debug output
next boot.
>
> For my problem at one place in the script "sigerr" was returned
by one
> command/function and incorrectly used as gateway ip address.
>
> Daniel
>
> I''ve just made another attempt.
Just after removing the claim_lock* lines, I also specified the
bridge in the file /etc/xen/xend-config.sxp as it follows:
(network-script ''network-bridge bridge=xenbr0'')

I always set up that line in the following way since so far:
(network-script network-bridge)

Now that I modified that line, I get the xenbr0 interface up at boot time,
but
there is also the peth0 interface (I don''t want it).

I simply had to perform a dhcpcd xenbr0 and the network came up.
But this should happen automatically, I don''t want to perform a dhcpcd
each time.
-- 
Flavio


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

> Just after removing the claim_lock* lines, I also specified the
> bridge in the file /etc/xen/xend-config.sxp as it follows:
> (network-script ''network-bridge bridge=xenbr0'')
I don''t know this syntax.
>
> I always set up that line in the following way since so far:
> (network-script network-bridge)
I do it this way. I made up a script "my-network-bridge" which calls
network-bridge with all my network cards.
>
> Now that I modified that line, I get the xenbr0 interface up at boot time,
> but
> there is also the peth0 interface (I don''t want it).
This is the physical interface located in dom0 which is in the bridge.
Before xen came up it was named eth0. After xen created xenbr0 it gets
renamed to peth0.
I do it the way that if I only want a bridge to interconnect domU''s
then I
use dummy network interfaces to build up a xenbrX of. Then I have pdummy0.
I don''t care about p*-interfaces.
>
> I simply had to perform a dhcpcd xenbr0 and the network came up.
> But this should happen automatically, I don''t want to perform a
dhcpcd
> each time.
I also had this kind of bug that the bridge wasn''t always upped
correctly
after xm startup.
I simply added "ifconfig xenbr0 0.0.0.0 up" to my
/etc/conf.d/local.start.

Daniel


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On 27 May 2010 12:55, Daniel Schmitt <daniel.schmitt@udo.edu> wrote:
> > Just after removing the claim_lock* lines, I also specified the
> > bridge in the file /etc/xen/xend-config.sxp as it follows:
> > (network-script ''network-bridge bridge=xenbr0'')
> I don''t know this syntax.
>
It''s described right in the configuration file /etc/xen/xend-config.sxp
.
There''s an example related to that configuration. So it''s
correct.
It simply explicitly says what is the bridge interface.

# The bridge is named xenbr0, by default.  To rename the bridge,
use
> #
> # (network-script ''network-bridge bridge=<name>'')
>

> I do it this way. I made up a script "my-network-bridge" which
calls
> network-bridge with all my network cards.
>
Mmmh :/ I''ve never done that. Actually I don''t know anything
about that.

> This is the physical interface located in dom0 which is in the bridge.
> Before xen came up it was named eth0. After xen created xenbr0 it gets
> renamed to peth0.
> I do it the way that if I only want a bridge to interconnect
domU''s then I
> use dummy network interfaces to build up a xenbrX of. Then I have pdummy0.
> I don''t care about p*-interfaces.
>
OK, thanks for the information.

> I also had this kind of bug that the bridge wasn''t always upped
correctly
> after xm startup.
> I simply added "ifconfig xenbr0 0.0.0.0 up" to my
/etc/conf.d/local.start.
>
I can try to do it on my system as well but I guess this is not sufficient
because of
the fact, the xend daemon could get up earlier. I''m gonna giving a try.

I will let you know.

Thanks a lot,

-- 
Flavio


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On 27 May 2010 17:17, Flavio <fbcyborg@gmail.com> wrote:
>  I also had this kind of bug that the bridge wasn''t always upped
correctly
>> after xm startup.
>> I simply added "ifconfig xenbr0 0.0.0.0 up" to my
/etc/conf.d/local.start.
>>
> I can try to do it on my system as well but I guess this is not sufficient
> because of
> the fact, the xend daemon could get up earlier. I''m gonna giving a
try.
>
Unfortunately, this didn''t fix my problem. The main reason is that the
local.start script
is the last one to be executed. So, xenbr0 bridge should start before the
xend daemon.
Even though the "net" is in the depend as it follows:
depend() {
    need net
    before xendomains sshd ntp-client ntpd nfs nfsmount rsyncd portmap dhcp
}

This is not sufficient to solve the problem.

I think it''s related to the xend init script, anyway.

-- 
Flavio


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Any news?

On 28 May 2010 00:33, Flavio <fbcyborg@gmail.com> wrote:
>
>
> On 27 May 2010 17:17, Flavio <fbcyborg@gmail.com> wrote:
>
>>  I also had this kind of bug that the bridge wasn''t always
upped correctly
>>> after xm startup.
>>> I simply added "ifconfig xenbr0 0.0.0.0 up" to my
>>> /etc/conf.d/local.start.
>>>
>> I can try to do it on my system as well but I guess this is not
sufficient
>> because of
>> the fact, the xend daemon could get up earlier. I''m gonna
giving a try.
>>
> Unfortunately, this didn''t fix my problem. The main reason is that
the
> local.start script
> is the last one to be executed. So, xenbr0 bridge should start before the
> xend daemon.
> Even though the "net" is in the depend as it follows:
> depend() {
>     need net
>     before xendomains sshd ntp-client ntpd nfs nfsmount rsyncd portmap dhcp
> }
>
> This is not sufficient to solve the problem.
>
> I think it''s related to the xend init script, anyway.
>
> --
> Flavio
>


-- 
Flavio


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Hello,

 > Any news?

I think I found the problem and a possible workaround:

in the function _claim_lock() in locking.sh a trap call on ERR is done 
(trap "release_lock $1; sigerr" ERR) which will trap all commands 
exiting with a non-zero status.
The problem is in the function do_ifup() in the network-bridge script: 
if no default gateway is found for the specified interface (e.g. because 
it''s not the main interface but only connected to some internal 
network), the statement ''[ -n "$gateway" ] && ip
route add default via
${gateway}'' will exit with non-zero code. This itself does not cause
the
trap (because it is inside a && statement) but the whole function will 
exit with the exit code of the last statement and thus ''do_ifup
$bridge''
(in op_start()) will return non-zero causing the trap...

A workaround is to substitute
[ -n "$gateway" ] && ip route add default via ${gateway}

with
if [ -n "$gateway" ]; then
             ip route add default via ${gateway}
fi

In previous xen releases (I don''t know exactly when they changed it) 
there were no calls to claim_lock and release_lock in the network-bridge 
script, so no error either :)

Does anyone know what the call to sigerr inside the trap command is for? 
As far as I know it is not a built-in bash command nor some common linux 
programm... that''s why the message "command not found" is
returned - not
a very helpful message though.

Katharina


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users