[Spread-users] spread abort after Deliver_packet: sequence error

martin345 at arcor.de martin345 at arcor.de
Thu Nov 16 05:18:59 EST 2006


Hi Jonathan, thanks for your response. My remarks included below. Thanks, Martin

----- Original Message ----
From:     Jonathan Stanton <jonathan at cnds.jhu.edu>
To:      martin345 at arcor.de
Date:   14.11.2006 10:31
Subject: Re: [Spread-users] spread abort after Deliver_packet: sequence error

> This error is a "must not occur" situation, which is why Spread kills itself
> with a fatal error message. So at some level it is definitely a Spread bug. 
> 
> There are two main sources for this type of error:
> 
> 1) a memory corruption problem, so the value stored in the field is
> corrupt.
> 2) a protocol bug where the spread code has a logic error.
> 
> Besides having high amounts of UDP traffic (I assume the extra traffic is
> not going to the Spread port numbers) is there anything else interesting about
> your setup?

Yes, the other UDP traffic is not going to the two spread ports.
Our spread setup has 8 hosts/segments and uses 43 groups. Nothing special in my opinion. Our spread.conf is attached to this mail. We do not use multicast.

> 
> Did you notice any other messages in teh Spread log that appeared unusual or
> occured in the last few seconds before the crash?

We observed a crash less than 4 hours after start. I attach the corresponding spread.log.
7 seconds before the crash there might be some suspicous messages:

2006-11-06 16:52:21 GMT Memb_token_loss: I lost my token, state is 1
2006-11-06 16:52:21 GMT Scast_alive: State is 2
2006-11-06 16:52:22 GMT Scast_alive: State is 2
2006-11-06 16:52:23 GMT Memb_handle_message: handling join message from -1726890574, State is 2
2006-11-06 16:52:23 GMT Send_join: State is 4
2006-11-06 16:52:23 GMT Memb_handle_message: handling join message from -1726890590, State is 4
... repeated ...
2006-11-06 16:52:27 GMT Memb_handle_message: handling join message from -1726890590, State is 4
...
2006-11-06 16:52:28 GMT Memb_handle_token: handling form2 token
2006-11-06 16:52:28 GMT Handle_form2 in FORM
2006-11-06 16:52:28 GMT Deliver_packet: sequence error: sec is 4, should be 1
Exit caused by Alarm(EXIT)

> 
> Does the crash occur often? If so how often?

This crash occurs not more than twice a week. Of course this is too much for us.

> 
> Thanks,
> 
> Jonathan
> 
> On Tue, Nov 07, 2006 at 11:51:41AM +0100, martin345 at arcor.de wrote:
> > spread-Version: 3.17.03
> > Platform: Solaris 2.9 with current patch cluster and 113459-04 SunOS 5.9:
> udp patch
> > 
> > A "Deliver_packet: sequence error" occured in spread.log (see below).
> Afterwards all spread daemons terminated itself.
> > 
> > Hint: We have heavy udp traffic on some hosts using spread because of SNMP
> polling, which is udp based like spread.
> > 
> > Can you give me any comments about that problem? Do you think there might
> be a problem inside spread?
> > 
> > Kind regards,
> > 
> > Martin
> > 
> > 2006-11-03 18:52:13 GMT Memb_handle_message: handling join message from
> -1726890574, State is 4
> > 2006-11-03 18:52:14 GMT Memb_handle_token: handling form2 token
> > 2006-11-03 18:52:14 GMT Handle_form2 in FORM
> > 2006-11-03 18:52:14 GMT Deliver_packet: sequence error: sec is -6, should
> be 2
> > Exit caused by Alarm(EXIT)
> > 
> > 
> > 
> > _______________________________________________
> > Spread-users mailing list
> > Spread-users at lists.spread.org
> > http://lists.spread.org/mailman/listinfo/spread-users
> 
> -- 
> -------------------------------------------------------
> Jonathan R. Stanton         jonathan at cs.jhu.edu
> Dept. of Computer Science   
> Johns Hopkins University    
> -------------------------------------------------------
> 






.

Viel oder wenig? Schnell oder langsam? Unbegrenzt surfen + telefonieren
ohne Zeit- und Volumenbegrenzung? DAS TOP ANGEBOT JETZT bei Arcor: günstig
und schnell mit DSL - das All-Inclusive-Paket für clevere Doppel-Sparer,
nur  44,85 €  inkl. DSL- und ISDN-Grundgebühr!
http://www.arcor.de/rd/emf-dsl-2




More information about the Spread-users mailing list