[Dxspider-support] Cluster Crashing

[Bob Harrison]

Thanks Dirk, I beat you too it by a few minutes :-)

I edited user_asc

deleted gb7baa

and rebuilt it with perl user_asc

I found that connect gb7baa crashed the cluster but so did sh/sta gb7baa!!!!

I figured it had fallen out with BAA.

Up and running again now

Cheers

bob

----- Original Message ----- 
From: "Dirk Koopman" <djk at tobit.co.uk>
To: "Bob Harrison" <bob at glh.demon.co.uk>
Sent: Thursday, October 18, 2007 8:31 PM
Subject: Re: Cluster Crashing


> Rebuild your user file
>
> stop the node
> cd /spider/data
> perl user_asc
> restart the node
>
> have a look at the sizes of user_asc and if the latest is huge then use 
> user_asc.o (.oo etc).
>
> I have been in Newcastle all day.
>
> Dirk
>
> Bob Harrison wrote:
>> Dirk,
>>
>> With some help from Ian VGS I have investigated more
>>
>> it appeared to crash on connects..... so I commented out the connects and 
>> GB7UJS started up ok and users connected to me.
>>
>> then I tried to do a manual connect and it fell over immediately.
>>
>> debug snip follows
>>
>>
>> My Crontab (after commenting out)
>> # Start connections to neighbouring clusters
>>
>> #
>>
>> #* * * * * start_connect('gb7baa') unless connected('gb7baa')
>>
>> #* * * * * start_connect('ir5pwj-6') unless connected('ir5pwj-6')
>>
>> #* * * * * start_connect('gb7ujs-2') unless connected('gb7ujs-2')
>>
>>
>>
>> Debug
>>
>> 1192733108^-> D G4UJS DX de SV1EQY: 7009.0 SV1AFW 1845Z
>>
>> 1192733109^<- I G4UJS connect gb7baa
>>
>> 1192733109^need a call and a reference at /spider/perl/LRU.pm line 52
>>
>> 1192733109^DXChannel GB7DXE destroyed (4)
>>
>> 1192733109^DXChannel ON4UN destroyed (3)
>>
>> 1192733109^DXChannel G4UJS destroyed (2)
>>
>> 1192733109^DXChannel GB7UJS destroyed (1)
>>
>> 1192733109^cluster
>>
>> HELP please
>>
>> ta
>>
>> Bob
>>
>>
>> Bob Harrison
>>
>> Telephone: +44 7876 791215
>>
>> Website: http://www.qsl.net/g4ujs
>