[Dxspider-support] PC41 loop?

Dirk Koopman G1TLH gb7tlh at dxcluster.org
Wed Jun 16 12:32:03 BST 2010


Rene

Thank you for reminding everyone of the reasons why the hop table 
exists. One could argue that it is no longer required. I am thinking 
that I might take it out or disable it - or maybe just use it for the 
"origination".

However, on the substantive points:-

1) hop count decrementing to 0: I believe that Lee's concerns were dealt 
with some time ago. AFAIK it now works as he believes it should. If it 
doesn't, then I would like some test cases to enable me to fix it.

2) There is a dup filter on PC41, it is currently set to a loop time of 
1 hour (3600 seconds). I can increase the default time to something else 
(24 hours?) or you can change it yourselves by doing a:

set/var $DXProt::eph_info_restime = 86400

I notice, taking just one example for today, that the loop time for 
these PC41 is huge (and the hop_table is not helping):

16Jun2010 at 03:49:29 <- I SM0RUX-6 PC41^IW9GXT^1^Daniele^H3^~
16Jun2010 at 05:02:42 <- I ED7ZAB-5 PC41^IW9GXT^1^Daniele^H2^~
16Jun2010 at 07:12:48 <- I IZ5FSA-6 PC41^IW9GXT^1^Daniele^H7^~
16Jun2010 at 11:20:31 <- I SM0RUX-6 PC41^IW9GXT^1^Daniele^H2^~

In the old days, 1 hour seemed to be enough to slug these messages, but 
that seems no longer to be the case. It may be that even one day is not 
long enough if these messages are getting re-hopped.

My concern is that, whatever loop constant we stick on here, there is 
some kind of slow background loop within the system as a whole that is 
causing this sort of problem.

In the limit there are three options:

1) switch them off (not at all my preferred solution, because the info 
is useful).
2) fiddle with the dedupe loop constant until they go away (enough).
3) add a new PC9x sentence that deals with it once and for all.

Dirk G1TLH


On 16/06/10 11:03, Rene Olsen wrote:
> Now it may be the case, that no cluser is no more than 10 hops away from each other. But
> that was not the case back then. The hops table was not introduced just for the fun of it.
>
> Some of the story also is that some cluster owners, thought that the spots from their node
> should not be able to propagate more than 5 nodes away. So they changed all hop counts to
> 5 or lower. Again, the hops table was not introduced just for the fun of it.
>
> It may well all be history now, but back in the days there were a cluster "war" here in Europe,
> where some siteops did everything in their power to prevent everything from spreading very
> far, and hence trying to force everyone to use their clusters. This is all history now, but the
> hops table was a way to make sure they didn't obtain their objective.
>
> Have done a bit of experimenting, and it seems that you can only raise the hop count and not
> decrease it. The reason that it can't be decreased, is of course to prevent owners being able
> to prevent things from being spread. And it doesn't seem to work on PC41 frames at all.
>
> I did another test to see how much PC41 trafic there is.
>
> My debug file 166.dat has a size of just above 62 MB. I then grep'ed all PC41 trafic from it,
> and the resulting file is very close to 17 MB !! That means that 27% of all the trafic is from
> PC41 frames. That is just insane imo. Something needs to be changed I think.
>
> Vy 73 de Rene / OZ1LQH
>
>
> On 16 Jun 2010 at 2:47, Lee Sawkins wrote:
>
>> We agree on what is happening.  You may not call it a bug, but I do.  The
>> reason hops must always be decremented and never increased is to prevent the
>> exact problem we are experiencing.  Other cluster node software does not
>> allow this "feature".
>>
>> A hops count of 15 or 20 should be enough to make it to every cluster in the
>> whole system.  As almost all clusters are now Internet linked, I doubt that
>> any cluster is now more than 10 hops away from any other cluster.
>>
>> Lee
>>
>> ----- Original Message -----
>> From: "Rene Olsen"<rene.chr.olsen at gmail.com>
>> To:<dxspider-support at dxcluster.org>
>> Sent: Wednesday, June 16, 2010 2:25 AM
>> Subject: Re: [Dxspider-support] PC41 loop?
>>
>>
>>> As far as I know there is no such thing as a hops bug in spider.
>>>
>>> But, it is a feature that has helped many in the past, that is now showing
>>> a bad side efect.
>>>
>>> Way back, when there were not as many clusters as there is now, the
>>> problem was that
>>> some clusters sent out the spots entered by their users with a hop count
>>> of 15 or 20.
>>>
>>> That hop count was just too low, for spots to make it all around, and it
>>> became obvious that
>>> some spots "died" along the way, because of the hop count.
>>>
>>> This is why the hops table was introduced into spider. With that file you
>>> can manipulate the
>>> hop count of every PC frame as you like. This is why you will often see a
>>> PC frame leave a
>>> spider node with a higher hop count than when it entered.
>>>
>>> Since we have dupe checking on spots and announcements, those gets
>>> blocked. But it
>>> seems like there is no dupe check for PC41 frames. But if all PC41 frames
>>> circles around the
>>> entire network 4 to 8 times a day, and maybe even for longer periods, then
>>> we are wasting a
>>> lot of bw, and I think this should be addressed. One way or another.
>>>
>>> Vy 73 de Rene / OZ1LQH
>>>
>>> On 16 Jun 2010 at 1:35, Lee Sawkins wrote:
>>>
>>>> This hops bug has been in Spider for years.  I started reporting it when
>>>> I
>>>> was a Spider sysop.  The problem is that Spider software does not
>>>> decrement
>>>> hops properly.
>>>>
>>>> I can send out PC messages from CC Clusters to Spider with a hops count
>>>> of
>>>> H1 and guess what, they come back to me with a higher hops count.  That
>>>> is
>>>> why I have turned PC41 off.   They just keep looping around.
>>>>
>>>> Lee
>>>>
>>>> ----- Original Message -----
>>>> From: "Rene Olsen"<rene.chr.olsen at gmail.com>
>>>> To:<dxspider-support at dxcluster.org>
>>>> Sent: Tuesday, June 15, 2010 11:59 PM
>>>> Subject: [Dxspider-support] PC41 loop?
>>>>
>>>>
>>>>> Hi.
>>>>>
>>>>> I was monitoring a bit with watchdbg yesterday and today, and started
>>>>> to
>>>>> wonder why there
>>>>> were so much PC41 trafic. I tehn decided to look at it a bit closer,
>>>>> because it looked to me,
>>>>> like it was the same information being sent over and over again.
>>>>>
>>>>> PC41^IZ7ECT^2^BARI^H16^~
>>>>>
>>>>> That one for example, I received no less than 6 times yesterday (debug
>>>>> file 166.dat).
>>>>>
>>>>> PC41^PA1LS^2^Halderberge^H3^~
>>>>>
>>>>> And that one I received 4 times.
>>>>>
>>>>> They don't come right after each other. The time stamps for the PA1LS
>>>>> one
>>>>> is:
>>>>>
>>>>> 1276560016
>>>>> 1276572511
>>>>> 1276620498
>>>>> 1276634794
>>>>>
>>>>> Seems to me that a lot of bw is wasted on having this stuff circling
>>>>> around.
>>>>>
>>>>> Is there any reason for this?
>>>>>
>>>>> I doubt that someone actually updates his QTH or other info several
>>>>> times
>>>>> a day :-)
>>>>>
>>>>> Vy 73 de Rene / OZ1LQH
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Dxspider-support mailing list
>>>>> Dxspider-support at dxcluster.org
>>>>> http://mailman.tobit.co.uk/mailman/listinfo/dxspider-support
>>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> Dxspider-support mailing list
>>> Dxspider-support at dxcluster.org
>>> http://mailman.tobit.co.uk/mailman/listinfo/dxspider-support
>>
>
>
>
>
> _______________________________________________
> Dxspider-support mailing list
> Dxspider-support at dxcluster.org
> http://mailman.tobit.co.uk/mailman/listinfo/dxspider-support
>




More information about the Dxspider-support mailing list