[Olsr-users] Received message to big to be forwarded

[Jernej Kos]

Hello!

Yes, I think that all the failing links were towards the central nodes
that were patched (the other nodes were not patched). But I can't be sure.

So on the failing nodes, all the routing table entries towards the
patched nodes were removed, while entries towards other neighbours were
still there.


Jernej

On 01. 05. 2015 10:15, Henning Rogge wrote:
> Hi,
> 
> just to be sure, we are talking about links between nodes that did not
> had the patch? Or links of other nodes to the "big central node" ?
> 
> Henning
> 
> On Thu, Apr 30, 2015 at 2:39 PM, Jernej Kos <(spam-protected)> wrote:
>> Hello!
>>
>> It seems I spoke too soon. After deploying the patch on two of the
>> largest nodes, several other nodes started gradually loosing
>> connectivity and after 4 hours around a hundred nodes were already
>> unreachable.
>>
>> Therefore I reverted the patch on those two nodes, which caused the
>> other nodes to become routable again. Really strange.
>>
>>
>> Jernej
>>
>> On 30. 04. 2015 11:33, Ferry Huberts wrote:
>>> I'll push it into deployment stream.
>>>
>>>
>>> On 30/04/15 11:23, Henning Rogge wrote:
>>>> Hi,
>>>>
>>>> I will push it, it should not prevent the aggregation of larger
>>>> packets, just the generation of large messages.
>>>>
>>>> We can give it a test until the next release... if we notice something
>>>> is very wrong with the message aggregation, I have to look over it
>>>> again.
>>>>
>>>> If it works, we can also backport it to the other release branches.
>>>>
>>>> Henning
>>>>
>>>>
>>>> On Thu, Apr 30, 2015 at 11:03 AM, Ferry Huberts <(spam-protected)>
>>>> wrote:
>>>>> Henning,
>>>>>
>>>>> Shall we integrate your patch?
>>>>> Can you push it?
>>>>>
>>>>> At least on master, unsure about other branches.
>>>>>
>>>>>
>>>>>
>>>>> On 30/04/15 10:44, Jernej Kos wrote:
>>>>>>
>>>>>> Hello!
>>>>>>
>>>>>> Sorry for taking so long, there were some other issues which prevented
>>>>>> the tests. I've tested this again today on the nodes with most links
>>>>>> and
>>>>>> it seems that the patch does fix the issue.
>>>>>>
>>>>>> I will push the patch to our OpenWrt feeds so that we can test it more.
>>>>>>
>>>>>>
>>>>>> Jernej
>>>>>>
>>>>>> On 28. 04. 2015 09:26, Henning Rogge wrote:
>>>>>>>
>>>>>>> Hi,
>>>>>>>
>>>>>>> any news about the test?
>>>>>>>
>>>>>>> Henning
>>>>>>>
>>>>>>> On Fri, Apr 17, 2015 at 7:33 PM, Jernej Kos <(spam-protected)> wrote:
>>>>>>>>
>>>>>>>> Hello!
>>>>>>>>
>>>>>>>> Ok, great, I will test it out. Thanks!
>>>>>>>>
>>>>>>>>
>>>>>>>> Jernej
>>>>>>>>
>>>>>>>> On 17. 04. 2015 19:32, Henning Rogge wrote:
>>>>>>>>>
>>>>>>>>> Yes,
>>>>>>>>>
>>>>>>>>> start with the "lots of neighbors" nodes... this should already
>>>>>>>>> resolve the problem. if it works (and if we cannot find a problem
>>>>>>>>> with
>>>>>>>>> it) we will commit it and it will be part of the next olsrd bugfix
>>>>>>>>> release.
>>>>>>>>>
>>>>>>>>> Henning
>>>>>>>>>
>>>>>>>>> On Fri, Apr 17, 2015 at 7:29 PM, Jernej Kos <(spam-protected)> wrote:
>>>>>>>>>>
>>>>>>>>>> Hello!
>>>>>>>>>>
>>>>>>>>>> So it will not help at all if we deploy it only at nodes
>>>>>>>>>> producing big
>>>>>>>>>> announces? Then this will not help much as we can't just upgrade
>>>>>>>>>> all
>>>>>>>>>> the
>>>>>>>>>> ~400 nodes at the moment.
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> Jernej
>>>>>>>>>>
>>>>>>>>>> On 17. 04. 2015 19:26, Henning Rogge wrote:
>>>>>>>>>>>
>>>>>>>>>>> Hi,
>>>>>>>>>>>
>>>>>>>>>>> you should deploy this patch on ALL nodes... it should work
>>>>>>>>>>> against
>>>>>>>>>>> the current "master branch"... but also against your source.
>>>>>>>>>>>
>>>>>>>>>>> Henning
>>>>>>>>>>>
>>>>>>>>>>> On Fri, Apr 17, 2015 at 7:25 PM, Jernej Kos <(spam-protected)> wrote:
>>>>>>>>>>>>
>>>>>>>>>>>> Hello!
>>>>>>>>>>>>
>>>>>>>>>>>> So if I understand this code correctly, this will limit the
>>>>>>>>>>>> maximum
>>>>>>>>>>>> message size that an originator node will produce? So I should
>>>>>>>>>>>> apply
>>>>>>>>>>>> this patch at the nodes with high neighbour count?
>>>>>>>>>>>>
>>>>>>>>>>>> Against which revision is this? I will test the patch over the
>>>>>>>>>>>> weekend
>>>>>>>>>>>> and report back.
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> Jernej
>>>>>>>>>>>>
>>>>>>>>>>>> On 17. 04. 2015 16:41, Henning Rogge wrote:
>>>>>>>>>>>>>
>>>>>>>>>>>>> Jernej,
>>>>>>>>>>>>>
>>>>>>>>>>>>> can you test the following patch?
>>>>>>>>>>>>>
>>>>>>>>>>>>> diff --git a/src/net_olsr.c b/src/net_olsr.c
>>>>>>>>>>>>> index afd24ec..4db54b1 100644
>>>>>>>>>>>>> --- a/src/net_olsr.c
>>>>>>>>>>>>> +++ b/src/net_olsr.c
>>>>>>>>>>>>> @@ -258,7 +258,14 @@ net_outbuffer_push_reserved(struct
>>>>>>>>>>>>> interface_olsr
>>>>>>>>>>>>> *ifp, const void *data, const
>>>>>>>>>>>>>    int
>>>>>>>>>>>>>    net_outbuffer_bytes_left(const struct interface_olsr *ifp)
>>>>>>>>>>>>>    {
>>>>>>>>>>>>> -  return ifp->netbuf.maxsize - ifp->netbuf.pending;
>>>>>>>>>>>>> +  /* IPv6 minimum MTU - IPv6 header - UDP header - VLAN-Tag */
>>>>>>>>>>>>> +  static int MAX_REMAINING = 1280 - 40 - 8 - 4;
>>>>>>>>>>>>> +  int remaining = ifp->netbuf.maxsize - ifp->netbuf.pending;
>>>>>>>>>>>>> +
>>>>>>>>>>>>> +  if (remaining > MAX_REMAINING) {
>>>>>>>>>>>>> +    return MAX_REMAINING;
>>>>>>>>>>>>> +  }
>>>>>>>>>>>>> +  return remaining;
>>>>>>>>>>>>>    }
>>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>> Henning
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>
>>>>>>
>>>>>
>>>>> --
>>>>> Ferry Huberts
>>>
>>

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: OpenPGP digital signature
URL: <http://lists.olsr.org/pipermail/olsr-users/attachments/20150503/a2d123cb/attachment.sig>