[Spread-users] Re: SP_scat_multicast not sending the entire message

Tudor Dumitras tdumitra at ece.cmu.edu
Tue Mar 1 14:18:23 EST 2005


Thanks a lot Taj, Daniel and John for all the information. This makes a 
lot of sense. In my tests, I don't think I was using messages larger 
than 64K. If I figure out how to reproduce that behavior, I'll post 
another message here.


Tudor


John Schultz wrote:
> Tudor Dumitras wrote:
> 
>> Daniel Essayag wrote:
>>
>>> Maybe your are testing with long messages. The max length is 10240 
>>> bytes.
>>
>>
>>
>> I'm sorry, I'm not familiar with this detail. What is this maximum 
>> length? This doesn't seem to be documented anywhere.
>>
>>
>> Tudor
>>
> 
> I believe the default maximum length Spread message is 102400 bytes, 
> which is an order of magnitude larger than what Daniel said.  However 
> this includes the size of some of the "headers" of a message, such as to 
> what groups you will send.  In general, I try to keep the data payload 
> portion of my msgs under 90K to ensure I don't bump into this limit.  I 
> I believe both the client library and daemon are doing checks to ensure 
> that the TOTAL size of a client send request is not larger than this.
> 
> If you attempted to send a message larger than 100K you should get a 
> failure return code and there should be no side effects (i.e. - no 
> partial msg sent, etc.).
> 
> If this limit is not covered in the documentation, we will ensure it is 
> added in the near future.
> 
> As to your original question Tudor, my best guess would be that some of 
> the code in the Spread client library isn't looping on (signal) 
> interruption as it should.  We will look into this as well.
> 






More information about the Spread-users mailing list