[Spread-users] Message counter wraparound bug

[Nico Meyer]

Hello John and others,

Is there any news regarding the message counter wraparound bug?

Creating an artificial partition with spmonitor works as a workaround, but 
this ia a real pain in the ass, since it is not scriptable and the time for 
the overflow to occour is reaching down to about 1 week for us now. If you 
just forget to 'reset' the counter once, then you're doomed.

I would consider this bug a serious showstopper for using spread in ciritical 
production systems. 
Is this bug really so complicated to fix? It is so critical, that I would have 
thought it to be high up on the priority list, so it must be hard and not 
just an obvious (to one of you at least) one line fix.
Perhaps you can give us (the community) a hint, how the message counter is 
supposed to work and where in the code this situation is handled. This part 
of the code is really hard to get into without an intimate knowledge of the 
protocol and the inner workings of Spread. You mentioned in an earlier mail 
that you made Spread open source, so people can contribute and fix bugs. But 
since there is no documentation on the Spread protocol, doing so needs a lot 
of dedication.

Hope to hear from you soon.

Cheers,

Nico Meyer