[Spread-users] 1225

[Jonathan Stanton]

On Mon, Apr 01, 2002 at 01:47:37PM -0800, Daniel Rall wrote:
> George Schlossnagle <george at omniti.com> writes:
> 
> > You can configure the maximum backlog at compile time (*cough* soft
> > limit *cough*), but ultimately, if you always send faster than you
> > recieve, you will fall behind and not be able to keep up.
> 
> Is there any plan of introducing such a soft limit?  CollabNet's QA
> just hit this 1000 message limit today while excercising some edges
> cases on a test box.  Our RPM would make it fairly easy to setup a
> compile-time change of a #define, but if someone can recommend a more
> forward-looking solution I would be willing to implement it.

Yes. I plan to add it to 3.17.0 which is the next release (after the 3.16.2
coming out today).

I was thinking of adding an option to spread.conf to set it at runtime.

Probably the monitor will also be able to change it during execution, but I
have to think and make sure there aren't any ways that could cause
deadlock/bad things. If the limit shrinks, we have to check it that causes
any connectsion to get cut off immediately. If the limit increases, then we
have to make sure to enable connections that now have room, but were blocked
before. 

Jonathan
-- 
-------------------------------------------------------
Jonathan R. Stanton         jonathan at cs.jhu.edu
Dept. of Computer Science   
Johns Hopkins University    
-------------------------------------------------------