[RC5] The recent crash.

simon at geordie.demon.co.uk simon at geordie.demon.co.uk
Tue Jun 23 09:58:28 EDT 1998

  I've had some thoughts about the recent crashes.
Surely just making the master key server bigger isn't a proper solution.
As Distributed.net grows, and the proxy network grows, the master server
would have to grow much faster to keep up.

Isn't a better way to make the proxies smarter. Either 
1) If a proxy can't find the key server, then it backs off a rendom amount
of time, if it finds the keyserver, but isn't getting blocks from it, then
it backs off a longer, random ammount of time. Hopefully if all the
proxies do this, then they won't all kit the key server at the same time
as soon as it's back. 


2) This is probably a bit harder to code. If a proxi can't find the
master, it should then try and contact other proxies. If it can find most
of them, and the other proxies also can't find the keyserver, then they
shoudl agree between them selves what order they should return their
blocks in when the keyserver comes back on line.  I.E. if there are 20
proxies, and they can't find the key server then they should agree between
them selves that each proxy should only try and contact the master for say
5 mins every houre ? Or they could agree a long enough slot to enable them
to flush their blocks. 

Something like that anyway.

Anyone any ideas ? Comments ? 

Please cc me all repplies as I only get the list in Digest format.

simon at geordie.demon.co.uk

To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest

More information about the rc5 mailing list