[B.A.T.M.A.N.] reboot sometimes doesn't work well

[B.A.T.M.A.N.] reboot sometimes doesn't work well

[Gioacchino Mazzurco]

Hi all!

Some times ago' in ninux Pisa we was experiencing kernel powering off
the machine caused by batman-adv, with ordex help we debugged what was
happening and the issue was solved bat that was possible because it was
happening on x86 computers ( with screen )

Now it SEEMS the same is happening but on mips ( Ubiquity M devices )
but here i don't know how to make sure the problem is this, the symptoms
is exactly the same, someone have idea how to save the powering off
kernel output on some file so after detach/attach cable we can find what
happened ?

Re: [B.A.T.M.A.N.] reboot sometimes doesn't work well

[Antonio Quartulli]

[-- Attachment #1: Type: text/plain, Size: 1049 bytes --]

Hello Gioacchino,

can you first tell us what version (and possibly which external patches) you are
using?

Also, can you remember the symptoms of the previous problem? I don't remember
very well what was happening...In this way everybody can understand what's
going on.

Cheers,

On Mon, Oct 01, 2012 at 01:24:40PM +0200, Gioacchino Mazzurco wrote:
> Hi all!
> 
> Some times ago' in ninux Pisa we was experiencing kernel powering off
> the machine caused by batman-adv, with ordex help we debugged what was
> happening and the issue was solved bat that was possible because it was
> happening on x86 computers ( with screen )
> 
> Now it SEEMS the same is happening but on mips ( Ubiquity M devices )
> but here i don't know how to make sure the problem is this, the symptoms
> is exactly the same, someone have idea how to save the powering off
> kernel output on some file so after detach/attach cable we can find what
> happened ?

-- 
Antonio Quartulli

..each of us alone is worth nothing..
Ernesto "Che" Guevara

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: [B.A.T.M.A.N.] reboot sometimes doesn't work well

[Gioacchino Mazzurco]

[-- Attachment #1: Type: text/plain, Size: 1290 bytes --]

On powering off kernel panicked reporting batman-adv tryng to access
some invalid pointer

Nor I remember exactly what was the solution but i remember you told me
that was about trying to doing something with an interface no longer exists

On 10/01/12 13:28, Antonio Quartulli wrote:
> Hello Gioacchino,
> 
> can you first tell us what version (and possibly which external patches) you are
> using?
> 
> Also, can you remember the symptoms of the previous problem? I don't remember
> very well what was happening...In this way everybody can understand what's
> going on.
> 
> Cheers,
> 
> On Mon, Oct 01, 2012 at 01:24:40PM +0200, Gioacchino Mazzurco wrote:
>> Hi all!
>>
>> Some times ago' in ninux Pisa we was experiencing kernel powering off
>> the machine caused by batman-adv, with ordex help we debugged what was
>> happening and the issue was solved bat that was possible because it was
>> happening on x86 computers ( with screen )
>>
>> Now it SEEMS the same is happening but on mips ( Ubiquity M devices )
>> but here i don't know how to make sure the problem is this, the symptoms
>> is exactly the same, someone have idea how to save the powering off
>> kernel output on some file so after detach/attach cable we can find what
>> happened ?
> 


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: [B.A.T.M.A.N.] reboot sometimes doesn't work well

[Gui Iribarren]

On Mon, Oct 1, 2012 at 8:24 AM, Gioacchino Mazzurco <gio@eigenlab.org> wrote:
> Hi all!
>
> Some times ago' in ninux Pisa we was experiencing kernel powering off
> the machine caused by batman-adv, with ordex help we debugged what was
> happening and the issue was solved bat that was possible because it was
> happening on x86 computers ( with screen )
>
> Now it SEEMS the same is happening but on mips ( Ubiquity M devices )
> but here i don't know how to make sure the problem is this, the symptoms
> is exactly the same, someone have idea how to save the powering off
> kernel output on some file so after detach/attach cable we can find what
> happened ?

Your best option is getting a serial cable; you *might* try in
/etc/config/system to enable persistent syslog output, and make it
write to the flash , but no guarantees you'll catch the kernel panic
that way. Serial console definitely would.

Does this happen every time?
> "on powering off"
you mean when you issue a "reboot" command through ssh, or what?
does it reboot succesfully, or simply hangs in there, requiring a
manual power cycle?

We haven't come across anything similar, using this hardware:
tl-mr3020, tl-mr3220, tl-mr3420, tl-wr842nd, ubnt bullet2 , ubnt bullet m2

Re: [B.A.T.M.A.N.] reboot sometimes doesn't work well

[Gioacchino Mazzurco]

reboot is called by a script called through ssh

> does it reboot succesfully,
no it doesn't

> or simply hangs in there, requiring a
> manual power cycle?
It hangs there but I lost connectivity the device became unreachable
from ipv4 and also ipv6, also link local ipv6 is gone :|


On 10/01/12 15:39, Gui Iribarren wrote:
> On Mon, Oct 1, 2012 at 8:24 AM, Gioacchino Mazzurco <gio@eigenlab.org> wrote:
>> Hi all!
>>
>> Some times ago' in ninux Pisa we was experiencing kernel powering off
>> the machine caused by batman-adv, with ordex help we debugged what was
>> happening and the issue was solved bat that was possible because it was
>> happening on x86 computers ( with screen )
>>
>> Now it SEEMS the same is happening but on mips ( Ubiquity M devices )
>> but here i don't know how to make sure the problem is this, the symptoms
>> is exactly the same, someone have idea how to save the powering off
>> kernel output on some file so after detach/attach cable we can find what
>> happened ?
> 
> Your best option is getting a serial cable; you *might* try in
> /etc/config/system to enable persistent syslog output, and make it
> write to the flash , but no guarantees you'll catch the kernel panic
> that way. Serial console definitely would.
> 
> Does this happen every time?
>> "on powering off"
> you mean when you issue a "reboot" command through ssh, or what?
> does it reboot succesfully, or simply hangs in there, requiring a
> manual power cycle?
> 
> We haven't come across anything similar, using this hardware:
> tl-mr3020, tl-mr3220, tl-mr3420, tl-wr842nd, ubnt bullet2 , ubnt bullet m2

Re: [B.A.T.M.A.N.] reboot sometimes doesn't work well

[Antonio Quartulli]

[-- Attachment #1: Type: text/plain, Size: 2078 bytes --]

On Mon, Oct 01, 2012 at 07:21:01PM +0200, Gioacchino Mazzurco wrote:
> reboot is called by a script called through ssh
> 
> > does it reboot succesfully,
> no it doesn't
> 
> > or simply hangs in there, requiring a
> > manual power cycle?
> It hangs there but I lost connectivity the device became unreachable
> from ipv4 and also ipv6, also link local ipv6 is gone :|

well, what about removing the batman-adv module _before_ rebooting? In this way
we can somehow move the problem.

(remember to do not connect via the mesh when you do this :-P)

ciao

> 
> 
> On 10/01/12 15:39, Gui Iribarren wrote:
> > On Mon, Oct 1, 2012 at 8:24 AM, Gioacchino Mazzurco <gio@eigenlab.org> wrote:
> >> Hi all!
> >>
> >> Some times ago' in ninux Pisa we was experiencing kernel powering off
> >> the machine caused by batman-adv, with ordex help we debugged what was
> >> happening and the issue was solved bat that was possible because it was
> >> happening on x86 computers ( with screen )
> >>
> >> Now it SEEMS the same is happening but on mips ( Ubiquity M devices )
> >> but here i don't know how to make sure the problem is this, the symptoms
> >> is exactly the same, someone have idea how to save the powering off
> >> kernel output on some file so after detach/attach cable we can find what
> >> happened ?
> > 
> > Your best option is getting a serial cable; you *might* try in
> > /etc/config/system to enable persistent syslog output, and make it
> > write to the flash , but no guarantees you'll catch the kernel panic
> > that way. Serial console definitely would.
> > 
> > Does this happen every time?
> >> "on powering off"
> > you mean when you issue a "reboot" command through ssh, or what?
> > does it reboot succesfully, or simply hangs in there, requiring a
> > manual power cycle?
> > 
> > We haven't come across anything similar, using this hardware:
> > tl-mr3020, tl-mr3220, tl-mr3420, tl-wr842nd, ubnt bullet2 , ubnt bullet m2

-- 
Antonio Quartulli

..each of us alone is worth nothing..
Ernesto "Che" Guevara

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: [B.A.T.M.A.N.] reboot sometimes doesn't work well

[Gioacchino Mazzurco]

It seems that compiling with newer openwrt revisions happily the problem
is not happening anymore so probably it was not batman-adv related

Sorry for the inconvenient but the behavior was pretty the same the
other time

Anyway thanks for debugging suggestions!

On 10/02/12 11:42, Antonio Quartulli wrote:
> On Mon, Oct 01, 2012 at 07:21:01PM +0200, Gioacchino Mazzurco wrote:
>> reboot is called by a script called through ssh
>>
>>> does it reboot succesfully,
>> no it doesn't
>>
>>> or simply hangs in there, requiring a
>>> manual power cycle?
>> It hangs there but I lost connectivity the device became unreachable
>> from ipv4 and also ipv6, also link local ipv6 is gone :|
> 
> well, what about removing the batman-adv module _before_ rebooting? In this way
> we can somehow move the problem.
> 
> (remember to do not connect via the mesh when you do this :-P)
> 
> ciao
> 
>>
>>
>> On 10/01/12 15:39, Gui Iribarren wrote:
>>> On Mon, Oct 1, 2012 at 8:24 AM, Gioacchino Mazzurco <gio@eigenlab.org> wrote:
>>>> Hi all!
>>>>
>>>> Some times ago' in ninux Pisa we was experiencing kernel powering off
>>>> the machine caused by batman-adv, with ordex help we debugged what was
>>>> happening and the issue was solved bat that was possible because it was
>>>> happening on x86 computers ( with screen )
>>>>
>>>> Now it SEEMS the same is happening but on mips ( Ubiquity M devices )
>>>> but here i don't know how to make sure the problem is this, the symptoms
>>>> is exactly the same, someone have idea how to save the powering off
>>>> kernel output on some file so after detach/attach cable we can find what
>>>> happened ?
>>>
>>> Your best option is getting a serial cable; you *might* try in
>>> /etc/config/system to enable persistent syslog output, and make it
>>> write to the flash , but no guarantees you'll catch the kernel panic
>>> that way. Serial console definitely would.
>>>
>>> Does this happen every time?
>>>> "on powering off"
>>> you mean when you issue a "reboot" command through ssh, or what?
>>> does it reboot succesfully, or simply hangs in there, requiring a
>>> manual power cycle?
>>>
>>> We haven't come across anything similar, using this hardware:
>>> tl-mr3020, tl-mr3220, tl-mr3420, tl-wr842nd, ubnt bullet2 , ubnt bullet m2
>