[Cluster-tech] "Incompatible Changes"

Dirk Koopman djk at tobit.co.uk
Mon Oct 8 20:45:49 CEST 2007


The changes I mentioned in a recent support posting today these:-

1. If the callsign after the 'A' or 'D' in a PC92 is the same as the 
node callsign then leave it out, so

	PC92^GB7RDX-1^66460^D^5GB7RDX-1^1N5IN-2^H91^

becomes

	PC92^GB7RDX-1^66460^D^^1N5IN-2^H91^

The same for A records.

2. I am going to change the time that C records are regenerated and sent 
back to 12 hours (initially) and hopefully to 24 hours eventually.

The heuristic for speeding up full network acquisition goes something 
like this:-

* on node connection (at both ends), send the current C record, then set 
the time before the next C record is regenerated and sent to 15 minutes. 
Reset the timer on every connection (this stops routing storms).

* I will investigate the concept of "if a new node appears out of the 
blue and I have not seen it 'recently' then regenerate and send a C 
record".

* this is for PC92 nodes only. External nodes will continue with the 
current 1 hour update, at least for now.

Dirk




More information about the Cluster-tech mailing list