[Spread-users] spread crash

[Jonathan Stanton]

That error occurs when something wrote outside of an allocated block of
memory and corrupted memory. 

The only identified case when this has occured in the past is when a
client process wrote 'too large' messages to Spread. Spread only supports
messages upto about 100kb in size. In some past versions the checks to
reject messages that were too large were not correct and large messages
could trigger memory corruption before being caught. I think the message
size checks are fixed in 3.16.1.

There might be some other way the corruption is getting caused. If you can
tell me a bit more about how you are running tests (what parameters to
flooder for example) I might see what the problem could be. 

We have run long flooder tests here and not had a problem, but our 'long'
is not as long as your 'long' :-) I.e. we've never run flooder for hours
continuously. Probably 5-10 minutes of so at full flooding is the most we
have done.

I do agree with the previous poster, to try 3.16.1. It fixes a number of
bugs in 3.16.0 and even if you havn't hit them yet, you probably will.

Jonathan

On Fri, Dec 14, 2001 at 05:37:03PM +0100, Abilio Oliveira wrote:
> 
>     Hi all:
> 
>     I'm using spread 3.16.0 and after running during two hours (and sometimes
> more than six or ten hours), with one sender spflooder under Linux Slackware and
> a receiver Flooder in java under  a Solaris system, the sender system fall down.
> The message is:
> 
> Membership id is ( -1408237526,1008232739)
> --------------------
> Configuration at Chatelet is:
> Num Segments 2
>         0       127.0.0.255     4803
>         2       172.16.0.255    4803
>                 Chatelet               172.16.0.42
>                 InsertoManage          172.16.0.50
> ====================
> ++++++++++++++++++++++
> Num of groups: 1
> [1] #Flooder#InsertoManage
> --------------
> spread: memory.c:608: dispose: Assertion `Mem_valid_objtype(obj_type)' failed.
> aborted
> 
> 
> Does anyone any one have any idea about this problem?
> By advance thanks for your responses. @bilio.
> 
> 
> _______________________________________________
> 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    
-------------------------------------------------------